Platformer

Why Platformer is leaving Substack

Casey Newton — Fri, 12 Jan 2024 01:00:49 GMT

“create a somber image displaying these characters: -30-” / DALL-E

After much consideration, we have decided to move Platformer off of Substack. Over the next few days, the publication will migrate to a new website powered by the nonprofit, open-source publishing platform Ghost. If you already subscribe to Platformer and wish to continue receiving it, you don’t need to do anything: your account will be ported over to the new platform.

If all goes well, following the Martin Luther King Jr. holiday on Monday, you’ll receive the Tuesday edition of Platformer as normal. If you have any issues with your subscription after that, please let us know.

Today let’s talk about how we came to this decision, the debate over how platforms should moderate content, and why we think we’re better off elsewhere.

When I launched Platformer on Substack in 2020, it was not in the belief that we would be here forever. Tech platforms come and go; in the meantime, they can also change in ways that make staying there impossible for the creators that rely on them. For this reason, I almost launched Platformer on a custom-built stack of services centered on WordPress, the way my inspiration Ben Thompson had done for Stratechery.

But Substack had some compelling advantages of its own. It was impressively fast and easy to set up. It paid to design Platformer’s logo. It offered me a year of healthcare subsidies, and ongoing legal support.

I also felt a personal connection to Substack’s co-founders, who believed that Platformer would succeed even before it had a name. They convinced me that I could thrive on their platform, and offered me a welcome boost in confidence as I considered leaving the best job I ever had to strike out on my own.

In the three years since, Substack has been a mostly happy home. Platformer has grown tremendously over that time, from around 24,000 free subscribers to more than 170,000 today. Our paid subscribers have allowed me to create new jobs in journalism. I’m proud of the work we do here.

Over that same period, Substack has faced occasional controversies over its laissez-faire approach to content moderation. The platform hosts a wide range of material I find distasteful and offensive. But for a time, the distribution of that material was limited to those who had signed up to receive it. In that respect, I did not view the decision to host Platformer on Substack as being substantially different from hosting it on, for example, GoDaddy.

But as I wrote earlier this week, Substack’s aspirations now go far beyond web hosting. It touts the value of its network of publications as a primary reason to use its product, and has built several tools to promote that network. It encourages writers to recommend other Substack publications. It sends out a weekly digest of publications for readers to consider subscribing to. And last year it launched a Twitter-like social network called Notes that highlights posts from around the network, regardless of whether you follow those writers or not.

Not all of you use these features. Some of you might not have seen them. But I can speak to their effectiveness: In 2023, we added more than 70,000 free subscribers. While I would love to credit that growth exclusively to our journalism and analysis, I believe we have seen firsthand how quickly and aggressively tools like these can grow a publication.

And if Substack can grow a publication like ours that quickly, it can grow other kinds of publications, too.

II.

In November, when Jonathan M. Katz published his article in The Atlantic about Nazis using Substack, it did not strike me as cause to immediately leave Substack. All platforms host problematic and harmful material; I assumed Substack would remove praise for Nazis under its existing policy that “Substack cannot be used to publish content or fund initiatives that incite violence based on protected classes.”

And so, after reading the open letter from 247 writers on the platform calling for clarity on the issue, I waited for a response.

The response, from Substack co-founder Hamish McKenzie, arrived on December 21. It stated that Substack would remove accounts if they made credible threats of violence but otherwise would not intervene. “We don't think that censorship (including through demonetizing publications) makes the problem go away — in fact, it makes it worse,” he wrote. “We believe that supporting individual rights and civil liberties while subjecting ideas to open discourse is the best way to strip bad ideas of their power.”

This was the moment where I started to think Platformer would need to leave Substack. I’m not aware of any major US consumer internet platform that does not explicitly ban praise for Nazi hate speech, much less one that welcomes them to set up shop and start selling subscriptions.

But suddenly, here we were.

I didn’t want to leave Substack without first getting my own sense of the problem. I reached out to journalists and experts in hate speech and asked them to share their own lists of Substack publications that, in their view, advanced extremist ideologies. With my colleagues Zoë Schiffer and Lindsey Choo, I reviewed them all and attempted to categorize them by size, ideology, and other characteristics.

In the end, we found seven that conveyed explicit support for 1930s German Nazis and called for violence against Jews, among other groups. Substack removed one before we sent it to them. The others we sent to the company in a spirit of inquiry: will you remove these clear-cut examples of pro-Nazi speech? The answer to that question was essential to helping us understand whether we could stay.

It was not, however, a comprehensive review of hate speech on the platform. And to my profound disappointment, before the company even acted on what we sent them, Substack shared the scope of our findings with another, friendlier publication on the platform, along with the information that these publications collectively had few subscribers and were not making money. (It later apologized to me for doing this.)

The point of this leak, I believe, was to make the entire discussion about hate speech on Nazis on Substack appear to be laughably small: a mountain made out of a molehill by bedwetting liberals.

To us, the six publications we had submitted had only ever been a question: would Substack, in the most clear-cut of all speech cases, do the bare minimum?

In the end, it did, in five out of six cases. As all of this unfolded, I spoke twice with Substack’s co-founders. And while they asked that those conversations be off the record, my understanding from our conversations — based on material they had shared with me in writing — was that in the future they would regard explicitly Nazi and pro-Holocaust material to be a violation of their existing policies.

But on Tuesday, when I wrote my story about the company’s decision to remove five publications, that language was missing from their statement. Instead, the company framed the entire discussion as having been about the handful of publications I had sent them for review.

I attempted to write a straightforward news story about all this, and wound up infuriating many readers. On the right, I faced criticism for making a fuss out of Substack hosting a handful of small Nazi publications. On the left, I faced even louder criticism for (in their view) appearing to celebrate and validate Substack’s removal of those same publications. (I wrote Tuesday that “Substack’s removal of Nazi publications resolves the primary concern we identified here last week.” I regret using that language. What I should have said was “Substack did the basic thing we asked it to,” and then emphasized that it did not address our larger concerns. Which I did go on to say, though not with the force that in hindsight I wish I had.)

I’m happy to take my lumps here. I just want to say again that to me, this was never about the fate of a few publications: it was about whether Substack would publicly commit to proactively removing pro-Nazi material. Up to the moment I published on Tuesday, I believed that the company planned to do this. But I no longer do.

From there, our next move seemed clear. But first I wanted to consult our readers, whose advice and support I have been so lucky to rely on over these past few years. Asking readers for their thoughts proved to be surprisingly controversial, especially in the Sidechannel Discord, where some of you wondered whether I was seeking a fig leaf of approval that we could use to justify staying here. But Platformer has as its readers some of the world’s smartest minds in content moderation and trust and safety — I sincerely wanted to get your thoughts before making a final decision.

Over the next 48 hours, the Platformer community raised a variety of sensible objections to how Substack had handled this issue. You pointed out that Substack had not changed its policy; that it did not commit explicitly to removing pro-Nazi material; that it seemed to be asking its own publications to serve as permanent volunteer moderators; and that in the meantime all of the hate speech on the platform remains eligible for promotion in Notes, its weekly email digest, and other algorithmically ranked surfaces.

In emails, comments, Substack Notes and callouts on social media, you’ve made your view clear: Platformer should leave Substack. We waited a day to announce our move as we finalized plans with Ghost and began our migration. But today we can say clearly that we agree with you.

Substack’s tools are designed to help publications grow quickly and make lots of money — money that is shared with Substack. That design demands responsible thinking about who will be promoted, and how.

The company’s defense boils down to the fact that nothing that bad has happened yet. But we have seen this movie before, from Alex Jones to anti-vaxxers to QAnon, and will not remain to watch it play out again.

III. Frequently asked questions about Substack and free speech

We’re still only talking about six newsletters. Aren’t you overreacting?

To be clear, there are a lot more than six bad publications on Substack: our analysis found dozens of far-right publications advocating for the great replacement theory and other violent ideologies.

But until Substack makes it clear that it will take proactive steps to remove hate speech and extremism, the current size of the problem isn’t relevant. The company’s edgelord branding ensures that the fringes will continue to arrive and set up shop, and its infrastructure creates the possibility that those publications will grow quickly. That’s what matters.

What about the fact that censorship will not make extremism go away, and might even make it worse?

We didn’t ask Substack to solve racism. We asked it to give us an easy, low-drama place to do business, and to commit to not funding and accelerating the growth of hate movements. Ultimately we did not get either.

Aren’t you actually helping Nazis here, making their ideology seem alluring by turning into forbidden knowledge?

Genocidal anti-semitism is hardly forbidden knowledge; you can find it just about anywhere. The Nazi worldview is taught in schools, figures prominently in popular culture, and endures in forums all across the internet. I believe Nazi ideology is made more appealing by the people who spread it than it is by the people who choose not to host it.

OK fine, but aren’t calls to ban Nazis a slippery slope? If Substack caves in here, there will be no end to what people like you call for them to remove.

The slippery-slope argument here is based on the fantasy that if you simply draw the right line, you will never have to revisit it. The fact is that we are constantly renegotiating the boundaries of speech based on values, norms, and threats of persecution. The problem is that renegotiating those boundaries is exhausting, expensive, and makes everyone mad, which is why most people prefer to shout “slippery slope!” and then never think about it again.

Previously you said you would submit your findings to Stripe, Substack’s payments processor. Doesn’t it set a bad precedent to lobby for content moderation at the level of payments?

Certainly I would not endorse a move like that in most cases. But as with Substack, I approached Stripe in the spirit of journalistic inquiry. One of its customers appeared to have said Nazis are free to set up shop there. Stripe’s policies forbid the funding of violent movements. How did Stripe’s policy square with Substack’s?

In any case, Stripe never responded to us. But it felt like it was worth sending an email.

Aren’t you going to have this exact same problem on Ghost, or wherever else you host your website?

As open-source software, Ghost is almost certainly used to publish a bunch of things we disagree with and find offensive. But it differs from Substack in some important respects.

One, its terms of service ban content that “is violent or threatening or promotes violence or actions that are threatening to any other person.” Ghost founder and CEO John O’Nolan committed to us that Ghost’s hosted service will remove pro-Nazi content, full stop. If nothing else, that’s further than Substack will go, and makes Ghost a better intermediate home for Platformer than our current one.

Two, Ghost tells us it has no plans to build the recommendation infrastructure Substack has. It does not seek to be a social network. Instead, it seeks only to build good, solid infrastructure for internet businesses. That means that even if Nazis were able to set up shop here, they would be denied access to the growth infrastructure that Substack provides them. Among other benefits, that means that there is nowhere on Ghost where their content will appear next to Platformer.

I actually reached these conclusions way before you did, and moved off Substack a long time ago. Is there anything you would like to say to me?

Good job! Thank you for your leadership.

IV.

I want to end on a note of gratitude.

We rarely see boom times in media, but these days the mood is particularly bleak. From smaller publications to giant ones, it can seem like almost everyone is laying people off and shrinking their ambitions. Against that backdrop, Substack has been a bright spot. Each week, some of the best things that I read are on the platform. It has made a meaningful, positive contribution to the culture, while also committing to allowing writers like me to leave when we want and take our customers with us when we do.

That takes talent, hard work, and principles. It counts for something. And I’m grateful to Substack for all the help it has given me and Platformer.

But over the past three years, another wonderful thing happened in media. Other publishing platforms sprung up featuring more robust content moderation policies, and with terms friendlier to business. (Platformer will save tens of thousands of dollars a year by no longer having to share 10 percent of its revenue with Substack.)

Meanwhile, as our paid readership grew into the thousands, we became less dependent on the goodwill of our host. So many times over the past two weeks, readers have written to say that they will follow Platformer anywhere it goes. Messages like yours have given us the confidence that, whatever challenges we might face as we leave this network, Platformer will endure.

Substack deserves credit for kicking off a revolution in independent publishing. But the world it helped to birth is now much bigger than its own platform. Next week we will move to a new home in that world. One where readers can feel confident their money is not going to accelerate the growth of hate movements. And one where we are no longer called upon constantly to defend an ideology we do not believe in.

I’m sad our time on Substack is ending in this way. But I’m also hopeful that, years from now, we will look back on today as a new beginning for our journalism.

And in the meantime, whatever else this decision turns out to mean for Platformer, I am certain in the knowledge that it is the right thing to do.

On the podcast this week: Kevin interrogates me about the decision to leave Substack. Then, the Wall Street Journal’s Kirsten Grind stops by to discuss her investigation into Elon Musk’s drug use. And finally, researcher Felix Wong joins us to explain how AI helped his team identify a new class of drugs that could potentially treat MRSA.

Talk to us

Send us tips, comments, questions, and questions about Ghost: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Meta seeks to hide harms from teens

Casey Newton — Wed, 10 Jan 2024 01:00:28 GMT

What teens will see as Meta rolls out changes to Instagram. (Meta)

Today, for a change of pace, let’s talk about something other than the fate of Substack and all those who dwell upon it.

Instead, let’s look at the mounting pressure on social networks to make their apps safer for young people, and the degree to which both platforms and regulators continue to talk past one another.

Let’s start with the news. Today, Meta said it would take additional steps to prevent users under the age of 18 from seeing content involving self-harm, graphic violence and eating disorders, among other harms.

Here’s Julie Jargon at the Wall Street Journal:

Teen accounts — that is, accounts of under-18 users, based on the birth date entered during sign-up — will automatically be placed into the most restrictive content settings. Teens under 16 won’t be shown sexually explicit content. On Instagram, this is called Sensitive Content Control, while on Facebook, it is known as Reduce. Previously, teens could choose less stringent settings. Teen users can’t opt out of these new settings.
The new restricted status of teen accounts means teens won’t be able to see or search for harmful content, even if it is shared by a friend or someone they follow. For example, if a teen’s friend had been posting about dieting, those posts will no longer be visible to the teen. However, teens might still see content related to a friend’s recovery from an eating disorder.

The changes announced today fall broadly into a category of platform tweaks that could be labeled “They weren’t doing that already?” But the fact that Meta made these changes, which the Journal calls “the biggest change the tech giant has made to ensure younger users have a more age-appropriate experience on its social-media sites,” underscores the degree to which child safety has become the most important dimension along which social networks are being judged in 2024.

How did that become the case?

The most consequential shift in attempts to regulate social media over the past year has been in the theory of harm. In the period after the 2016 US presidential election, regulators focused heavily on issues related to speech. Democrats focused on the way platforms can amplify lawful but harmful material, including hate speech and misinformation. Republicans took the opposite position, protesting against platforms’ right to moderate content and calling for an end to account removals and other restrictions in most cases.

The result was an enervating stalemate in efforts to regulate US tech companies; Congress has not passed a single meaningful new tech regulation since Donald Trump was elected president. But after the better part of a decade waiting for lawmakers to act, states took matters into their own hands. And in some cases, the focus remained on speech, as in the laws passed by Florida and Texas aimed at making content moderation illegal.

But over the past year, states have adopted a new approach to reducing the growth and power of social networks: making it increasingly difficult for children to use them. Utah and Arkansas have both passed laws seeking to ban minors from using some social networks without their parents’ permission. Montana sought to ban TikTok entirely.

Laws like these have struggled so far in courts. A lawsuit is seeking to block the Utah legislation from being implemented; and federal judges blocked both the Arkansas law and the Montana law from taking effect.

At the same time, pressure for lawmakers to act in these cases has only grown. Last year, the US surgeon general issued an advisory warning that social networks can be harmful to teens. Some critics dismissed the report as political posturing, but on the whole I thought it offered a fairly balanced portrait of both the potential benefits and harms that can come with extended use of social networks.

Meanwhile, a series of investigations over the past year has continued to shine a light on ways that social networks continue to be unsafe for young people. The Wall Street Journal documented how Instagram is used to connect buyers and sellers of child sexual abuse material, and the Stanford Internet Observatory found that similar issues related to CSAM affect X, Telegram, Mastodon, and other decentralized networks.

More recently, the Journal reported that Meta’s systems are unwittingly connecting pedophiles together. And on Snapchat, a scam targeting teenage boys has resulted in more than a dozen children taking their lives after being blackmailed over their nudes.

Frustration with platforms culminated in the lawsuit filed against Meta in October by 41 states and the District of Columbia, alleging that the company misled young people about the potential harms they face on the company’s apps. An unredacted version of the suit that came out later documented the extent to which Meta was aware of huge numbers of under-13 users.

All of that feels like necessary context for the changes Meta announced today. It also feels like an explanation why, despite the changes it has made over the past year, Meta remains on the defensive.

Last July, after he signed the Utah legislation, Gov. Spencer Cox described his rationale in an interview with the New York Times’ Jane Coaston. “If you look at the increased rates of depression, anxiety, self-harm since about 2012, across the board but especially with young women, we have just seen exponential increases in those mental health concerns,” Cox told her. “Again, the research is telling us over and over and over again that it is not just correlated, but it’s being caused, at least in part, by the social media platforms.”

That view is much disputed by the social networks, which argue that the data is more muddled and that any causal effects are small to nonexistent. But Cox’s belief is widely shared — and not just among Republican governors, but among millions of families who have complicated relationships with Facebook, Instagram, YouTube, TikTok, Snapchat, and other apps.

If you believe that at least some teens experience harm on social media, it seems unlikely that a parental permission slip will solve it. Neither, I’m afraid, will a few tweaks to where content about self-harm can be shared.

Not for the first time, lawmakers and social networks are talking past each other. To make real progress on teen mental health in 2024, both sides are going to have to find some common ground.

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

TikTok quietly restricted a data analysis tool intended to support advertisers after researchers used it to study content related to the Israel-Hamas war, among other topics. (Sapna Maheshwari / The New York Times)
X banned a number of high-profile accounts, including journalists, podcasters and writers that criticized the Israeli government. Their bans were quickly overturned, with Elon Musk blaming the move on unruly spam algorithms. (Thomas Germain / Gizmodo)
The European Commission is examining Microsoft and OpenAI’s relationship to consider whether it needs to be vetted under merger rules. (Samuel Stolton / Bloomberg)
- A profile of Anna Makanju, OpenAI’s vice president of global affairs, who helped position the company as a trusted partner for lawmakers. (Cat Zakrzewski / Washington Post)
Large language models from Meta and OpenAI have spawned a wave of digital sex bots — some of whom advertise themselves as children. (Ben Weiss and Alexandra Sternlicht / Fortune)
Apple is pushing back against European rules that designate its five App Stores as one core platform service subject to regulation, saying that each store is specific to its devices. Uh huh. (Foo Yun Chee / Reuters)
A Chinese state-backed institute says it has developed a technique to identify people who use Apple AirDrop to send messages. A potentially worrisome blow to protesters. (Bloomberg)
Elections around the world are facing a new spread of disinformation amid rising extremism, rapid AI advancements, and a dangerous lack of content moderation. (Tiffany Hsu, Stuart A. Thompson and Steven Lee Myers / The New York Times)
UK media regulator Ofcom has a new online safety team of almost 350 people, including new senior hires from Meta, Microsoft, and Google. (Cristina Criddle / Financial Times)
The Screen Actors Guild signed a deal requiring consent and minimum payments when actors’ voices are digitally replicated. (Thomas Buckley / Bloomberg)

Industry

ByteDance is in talks with multiple companies, including gaming giant Tencent, to sell off its gaming assets. (Josh Ye / Reuters)
Volkswagen is planning on installing ChatGPT in cars, rolling them out first in Europe then in the US. ??? (Andrew J. Hawkins / The Verge)
A number of AI-generated ads are impersonating celebrities to endorse products in a scarily accurate way. So no, the Taylor Swift Le Creuset giveaway ad is not real. (Tiffany Hsu and Yiwen Lu / The New York Times)
Startup founders are worried that they might get “Altmaned” — ousted by their own company — and are looking at ways to protect themselves. (Corrie Driebusch / The Wall Street Journal)
Activist investor Elliott Investment Management invested about $1 billion in Match, the parent company of Tinder. Remember when Elliott did this with Jack Dorsey’s Twitter? I forget how that turned out. (Lauren Thomas / The Wall Street Journal)
Dating apps like Tinder and Hinge are raising prices and introducing new subscription tiers, testing how much users will pay to find love. (Euan Healy / Financial Times)
X’s former head of trust and safety, Ella Irwin, joined Stability AI as its senior vice president of integrity. (Ben Goggin / NBC News)
X announced a number of content deals for streaming, including shows hosted by former CNN anchor Don Lemon and former Fox Sports host Jim Rome. (Alex Weprin / The Hollywood Reporter)
Demos for the Apple Vision Pro will begin in Apple Stores at the start of February. (Zac Hall / 9to5Mac)
Apple is asking developers not to refer to visionOS apps as “AR” or “VR”, under its new guidelines. Not sure even Apple can make “spatial computing” the standard parlance. (Filipe Espósito / 9to5Mac)
Carta is closing its liquidity services business, following allegations that a salesperson from the department improperly used customer data. (Dan Primack and Kia Kokalitcheva / Axios)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and teen mental health solutions: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Substack says it will remove Nazi publications from the platform

Casey Newton — Mon, 08 Jan 2024 22:49:21 GMT

(Rafael Henrique / Getty Images)

Substack is removing some publications that express support for Nazis, the company said today. The company said this did not represent a reversal of its previous stance, but rather the result of reconsidering how it interprets its existing policies.

As part of the move, the company is also terminating the accounts of several publications that endorse Nazi ideology and that Platformer flagged to the company for review last week.

The company will not change the text of its content policy, it says, and its new policy interpretation will not include proactively removing content related to neo-Nazis and far-right extremism. But Substack will continue to remove any material that includes “credible threats of physical harm,” it said.

In a statement, Substack’s co-founders told Platformer:

If and when we become aware of other content that violates our guidelines, we will take appropriate action.
Relatedly, we’ve heard your feedback about Substack’s content moderation approach, and we understand your concerns and those of some other writers on the platform. We sincerely regret how this controversy has affected writers on Substack.
We appreciate the input from everyone. Writers are the backbone of Substack and we take this feedback very seriously. We are actively working on more reporting tools that can be used to flag content that potentially violates our guidelines, and we will continue working on tools for user moderation so Substack users can set and refine the terms of their own experience on the platform.

Substack’s statement comes after weeks of controversy related to the company’s mostly laissez-faire approach to content moderation.

In November, Jonathan M. Katz published an article in The Atlantic titled “Substack Has a Nazi Problem.” In it, he reported that he had identified at least 16 newsletters that depicted overt Nazi symbols, and dozens more devoted to far-right extremism.

Last month, 247 Substack writers issued an open letter asking the company to clarify its policies. The company responded on December 21, when Substack co-founder published a blog post arguing that “censorship” of Nazi publications would only make extremism worse.

McKenzie also wrote that “we don’t like Nazis either” and said Substack wished “no-one held those views.” But “we don't think that censorship (including through demonetizing publications) makes the problem go away,” he wrote. “In fact, it makes it worse. We believe that supporting individual rights and civil liberties while subjecting ideas to open discourse is the best way to strip bad ideas of their power.”

The statement seemed to be at odds with Substack’s published content guidelines, which state that “Substack cannot be used to publish content or fund initiatives that incite violence based on protected classes.”

In its aftermath, several publications left the platform. Others, including Platformer, said they would leave if the company did not remove pro-Nazi publications.

Meanwhile, more than 100 other Substack writers, including prominent names like Bari Weiss and Richard Dawkins, signed a post from writer Elle Griffin calling on Substack to continue with its mostly hands-off approach to platform-level moderation.

From its inception, McKenzie and Substack co-founder Chris Best have touted freedom of speech as one of Substack’s core virtues. As a result, the platform has been embraced by fringe thinkers, who have built large businesses while promoting anti-vaccine pseudo-science, Covid conspiracy theories and other material that is generally restricted on mainstream social networks.

Substack has defended its approach by arguing that it is built differently from social networks, which optimize for engagement rather than subscription revenue. The company says it employs a “decentralized” approach to moderation that allows individual readers to decide which writers they want to subscribe to; and lets writers determine which comments they will allow and which blogs they will recommend.

(Incidentally, this approach means that you can’t currently report comments directly to Substack: only writers receive your reports. Platformer has reviewed several cases of violent material and death threats in Substack comments.)

At the same time, over the past couple years Substack has come to more closely resemble the social networks it often criticizes. Each week, Substack sends users a personalized, algorithmically ranked digest of posts from writers they don’t yet follow — a feature that can help fringe publications build larger audiences and make more money than they would otherwise.

And last year Substack launched Notes, a text-based social feed similar to Twitter that also surfaces personalized content in a ranked feed. Notes can also give heightened visibility and free promotion to extremists.

The question now is whether taking action against some pro-Nazi accounts will shift the perception that Substack is a home for the most extreme ideologies, and prevent an exodus among writers who prefer more aggressive content moderation.

In recent weeks, Platformer has worked with other journalists and extremism researchers in an effort to understand the scope of far-right content on the platform. We’ve now reviewed dozens of active, monetized publications that advance violent ideologies, including anti-Semitism and the great replacement theory.

Substack has argued that extremist publications represent only a small fraction of newsletters on the platform, and as far as we can tell this is true. At the same time, the site’s recommendations and social networking infrastructure is designed to enable individual publications to grow quickly. And the company’s outspoken embrace of fringe viewpoints all but ensures that the number of extremist publications on the platform will grow.

The company is now in a difficult position. Having branded itself as a bastion of free speech, any changes to its content policy risks driving away writers who chose the platform in part for its rejection of aggressive content moderation. At the same time, other publications — Platformer included — have lost scores of paying customers who do not want to contribute to a platform that they see as advancing the cause of extremism.

In coming days, explicitly Nazi publications on Substack are slated to disappear. But the greater divide within its user base over content moderation will remain. The next time the company has a content moderation controversy — and it will — expect these tensions to surface again.

What this means for Platformer

Substack’s removal of Nazi publications resolves the primary concern we identified here last week. At the same time, as noted above, this issue has raised concerns that go beyond the small group of publications that violate the company’s existing policy guidelines.

As we think through our next steps, we want to hear from you. If you have unsubscribed from Platformer or other publications over the Nazi issue, does the company’s new stance resolve your concerns? Or would it take more? If so, what?

Paid subscribers can comment below; everyone is welcome to email us with their thoughts.

Investors are focused on these metrics.

Startups should take notice.

It takes more than a great idea to make your ambitions real. That’s why Mercury goes beyond banking* to share the knowledge and network startups need to succeed. In this article, they shed light on the key metrics investors have their sights set on right now.

Even in today’s challenging market, investments in early-stage startups are still being made. That’s because VCs and investors haven’t stopped looking for opportunities — they’ve simply shifted what they are searching for. By understanding investors’ key metrics, early-stage startups can laser-focus their next investor pitch to land the funding necessary to take their company to the next stage.

Read the full article to learn how investors think and how you can lean into these numbers today.

Read the Article

*Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank & Trust®; Members FDIC.

Platformer has been a Mercury customer since 2020. This sponsorship gets us 5% closer to our goal of hiring a reporter in 2024.

Governing

The US Department of Justice is in the late stages of investigating Apple’s strategies to protect its iPhone dominance. A “sweeping” antitrust case may be forthcoming. (David McCabe and Tripp Mickle / The New York Times)
YouTube updated its harassment policies to prohibit deepfakes that depict minors or victims of crimes that are describing the violence they experienced. Catching up to a disturbing trend on video platforms. (Mia Sato / The Verge)
OpenAI published its rebuttal to the New York Times lawsuit, saying the paper’s claims “without merit.” They go particularly hard on claims that ChatGPT routinely regurgitates copyrighted content, calling it a rare bug. (OpenAI)
Arizona is using AI to test some worst-case Election Day scenarios in an attempt to dispel AI-generated misinformation. Here’s wishing AI luck in its fight against AI. (NPR Morning Edition)
The Supreme Court rejected a request by X to consider whether the platform can reveal how often federal agents ask for user information as part of their national security investigations. (Nate Raymond / Reuters)
The big three cloud computing providers — Amazon, Microsoft and Google — offer only narrow legal indemnities against AI copyright claims against their users, legal experts say. (Camilla Hodgson / Financial Times)
- Also: AI image generators from companies like Midjourney and OpenAI should be limited to properly licensed content to avoid plagiarism, these authors argue. (Gary Marcus and Reid Southen / IEEE Spectrum)
European antitrust head Margarethe Vestager is set to meet with the chief executives of Apple, Alphabet, Broadcom and Nvidia this week, along with executives from OpenAI. (Foo Yun Chee / Reuters)
The European Commission is urging Google and other tech companies to promote stories by Belarusian journalists opposing the country’s regime, after complaints that pro-regime stories are being pushed by algorithms. (Raphael Minder / Financial Times)
Amazon, Alphabet and Microsoft are among the tech giants boosting their Saudi Arabian presence, after authorities there said they would stop giving contracts to companies that lack a regional headquarters. Gross! (Matthew Martin and Fahad Abuljadayel / Bloomberg)

Industry

Microsoft executive Dee Templeton has reportedly joined OpenAI’s board as a nonvoting observer. (Dina Bass and Rachel Metz / Bloomberg)
Threads is taking steps to address “low-quality recommendations”, according to Instagram head Adam Mosseri. The statement came after users began seeing lots of anti-LGBT posts in their recommendations. (Jay Peters / The Verge)
The Twitch “clips” feature — one of the least moderated aspects the site — is being used to record and share child sexual abuse material, an analysis found. (Cecilia D’Anastasio / Bloomberg)
How the desperate race for Google Search traffic shapes websites and impacts content, often at the expense of real people. An excellent look at how Google warps the entire shape of the web with their content guidelines. (Mia Sato / The Verge)
TikTok is partnering with Peloton to create the #TikTokFitness hub, which includes live Peloton classes, class clips and original instructor series. (Dean Seal / The Wall Street Journal)
X continues to have a crypto ad scam problem, with the number of malicious ads increasing rapidly over the past month, according to researchers. (Lawrence Abrams / Bleeping Computer)
Leaders at Elon Musk’s companies are reportedly concerned about Musk’s use of illegal drugs like LSD, cocaine and ecstasy. (Emily Glazer and Kirsten Grind / The Wall Street Journal)
Google seems to be working on Bard Advanced, an upgraded version of Bard that will be available through paid subscriptions to Google One. (Emma Roth / The Verge)
Apple’s Vision Pro VR headset is set to ship out to customers in February, with pre-orders beginning on Jan. 19. (Emma Roth / The Verge)
Spotify’s shift toward AI recommendations means that once-influential, human-curated playlists like Rap Caviar are losing their power. They’re being streamed less, too. (Ashley Carman / Bloomberg)
Midjourney v6, the newest image synthesis model by the AI image company, has massive improvements in detail and scenery, but its images still suffer from high contrast and saturation. Plus, it still scrapes artwork by artists without their explicit consent. (Benj Edwards / Ars Technica)
Carta’s CEO says the startup is investigating potential misuse of customer data, following allegations that it had used its knowledge of investors to promote a Linear share sale. The story everyone in Silicon Valley was talking about over teh weekend.(Anne VanderMey / Bloomberg)
AI hallucinations should be appreciated for their creativity, this author argues. They also highlight the importance of human fact-checking and decision-making. (Steven Levy / WIRED)
LLMs like ChatGPT perform better when assigned gender-neutral roles, researchers found. (Sisi Wei / The Markup)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and your feedback on Substack’s Nazi enforcement: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Why Substack is at a crossroads

Casey Newton — Fri, 05 Jan 2024 01:27:38 GMT

(Rafael Henrique / Getty Images)

Most days, this column looks at controversies unfolding on other tech platforms. Today, let’s take a look at the one that hosts this publication: Substack.

On Tuesday, I told subscribers that we are considering leaving the platform based on the company’s recent statement that it would not demonetize or remove openly Nazi accounts. After Jonathan M. Katz’s November article investigating extremism on the platform in The Atlantic, 247 Substack writers published an open letter asking the company to clarify its policies.

A few days later, Substack co-founder Hamish McKenzie responded in a blog post. While the platform would remove publications that are found to make credible threats of violence — a high bar — Substack would otherwise leave them alone, he said. “We don't think that censorship (including through demonetizing publications) makes the problem go away — in fact, it makes it worse,” he wrote. “We believe that supporting individual rights and civil liberties while subjecting ideas to open discourse is the best way to strip bad ideas of their power.”

McKenzie’s perspective — that sunlight is the best disinfectant, and that censorship backfires by making dangerous ideas seem more appealing — is reasonable for many or even most circumstances. It is a point of view that informs policies at many younger, smaller tech platforms, owing both to the techno-libertarian streak that runs through many founders in Silicon Valley and the fact that a hands-off approach to content moderation is easier and less expensive than the alternatives.

There was a time when even Facebook, which has more restrictive policies than Substack does across the board, permitted users to deny the Holocaust. CEO Mark Zuckerberg occasionally cited this policy as evidence of the company’s commitment to free speech, even though it occasionally got him into trouble.

Then, in 2020, Facebook reversed course: going forward, it said, it would remove Holocaust denial from the platform. In doing so, Zuckerberg said, Facebook was seeking to keep pace with the changing times.

Here’s Sheera Frenkel in the New York Times:

In announcing the change, Facebook cited a recent survey that found that nearly a quarter of American adults ages 18 to 39 said they believed the Holocaust either was a myth or was exaggerated, or they weren’t sure whether it happened.
“I’ve struggled with the tension between standing for free expression and the harm caused by minimizing or denying the horror of the Holocaust,” Mr. Zuckerberg wrote in his blog post. “Drawing the right lines between what is and isn’t acceptable speech isn’t straightforward, but with the current state of the world, I believe this is the right balance.”

For a time when memories of the Holocaust were fresh, and anti-Semitism had ebbed, it may have seemed less dangerous to let a few cranks peddle their lies. But as those memories faded, and attacks on Jewish people surged, Facebook felt compelled to revisit those policies.

That brings us to Substack. When it was founded in 2017, Substack offered simple infrastructure for individuals to create and grow their email newsletters. From the start, it promised not to take a heavy hand with content moderation. And because it only offered software, this approach drew little criticism. If you wrote something truly awful in Word, after all, no one would blame Microsoft. Substack benefited similarly from this distance.

Over time, though, the company evolved. It began encouraging individual writers to recommend one another, funneling tens of thousands of subscribers to like-minded people. It started to send out an algorithmically ranked digest of potentially interesting posts to anyone with a Substack account, showcasing new voices from across the network. And in April of this year, the company launched Notes, a text-based social network resembling Twitter that surfaces posts in a ranked feed.

By 2023, in other words, Substack no longer could claim to be the simple infrastructure it once was. It was a platform: a network of users, promoted via a variety of ranked surfaces. The fact that it monetized through subscriptions rather than advertising did not change the fact that, just as social networks have at times offered unwitting support to extremists, Substack was now at risk of doing the same.

And in one key respect, Substack is even more vulnerable to this criticism than social networks had been. Extremists on Facebook, Twitter, and YouTube for the most part had been posting for clout: those platforms made it difficult or even impossible for them to monetize their audiences.

On Substack, on the other hand, extremists can post for money. The pieces are now all in place for an extremist Substack to grow an audience using the platform’s recommendation systems, and monetize that audience via subscriptions. And Substack, as it does with all publications, will get 10 percent of the revenue.

Now, other platforms have developed defenses against their systems being exploited in this way. Some, like Facebook, prevent designated dangerous organizations from starting accounts. They also might ban praise for those organizations, or ban hate speech in general. Others, like YouTube, might allow some speech that makes the company uncomfortable, but restrict them from monetizing. Or it will restrict them from appearing in search, or in recommendations.

Substack doesn’t want to do that. It wants to be seen as a pure infrastructure provider — something like Cloudflare, which seemingly only has to moderate content once every few years. But Cloudflare doesn’t recommend blogs. It does not send out a digest of websites to visit. It doesn’t run a text-based social network, or recommend posts you might like right at the top.

Recommendations might appear on their surface to be innocuous, and in most cases they are. In three years on Substack, I’ve been recommended plenty of boring posts, but no openly Nazi ones. My experience of them has been unobjectionable.

But turning a blind eye to recommended content almost always comes back to bite a platform. It was recommendations on Twitter, Facebook, and YouTube that helped turn Alex Jones from a fringe conspiracy theorist into a juggernaut that could terrorize families out of their homes. It was recommendations that turned QAnon from loopy trolling on 4Chan into a violent national movement. It was recommendations that helped to build the modern anti-vaccine movement.

The moment a platform begins to recommend content is the moment it can no longer claim to be simple software.

It is, of course, exhausting — and expensive — to have to police your platform this way. Some users really do want to censor everyone who disagrees with them, and lobby to remove all of their political opponents from the platform. Finding the real danger and harms in a sea of user reports is tedious, thankless work. And no matter how you choose to moderate content, you’ll make at least some groups of users mad.

At the same time, for the sake of your business, you have to draw a line somewhere.

Some of these lines are quite tricky — do you ban all nudity? What if a mother is breastfeeding?

Others are not. Until Substack, I was not aware of any major US consumer internet platform that stated it would not remove or even demonetize Nazi accounts. Even in a polarized world, there remains broad agreement that the slaughter of 6 million Jews during the Holocaust was an atrocity. The Nazis did not commit the only atrocity in history, but a platform that declines to remove their supporters is telling you something important about itself.

If it won’t remove the Nazis, why should we expect the platform to remove any other harm?

Our readers understand this. During the past couple weeks, dozens of paid subscribers to Platformer have canceled their memberships. “The reason is simple,” one of those readers wrote to us today. “I don't want to fund Nazis. I'm disturbed by a Substack leadership that looks at openly pro-Nazi content and says, ‘We won't de-platform you. In fact, we'll monetize you.’"

I’m proud of the Platformer readership for standing up for their principles in this way. Some of our earliest and best customers are people who work in tech policy, content moderation, and trust and safety. They’ve spent years doing the work, making the hard calls, and cleaning up the internet for all of our mutual benefit. It’s only natural that they would resist spending money on a platform that spurns their profession in this way.

Over the past few days, the Platformer team analyzed dozens of Substacks for pro-Nazi content. Earlier this week, I met with Substack to press my case that they should remove content that praises Nazis from the network. Late today, we submitted a list of accounts that we believe to be in violation of the company’s existing policies against incitement to violence. I am scheduled to meet with the company again tomorrow.

Whatever becomes of those accounts, though, I fully expect that more will spring up in their wake. So long as Substack allows itself to be perceived — encourages itself to be perceived! — as a home for Nazis, they will open accounts here and start selling subscriptions. Why wouldn’t they?

Every platform hosts its share of racists, white nationalists, and other noxious personalities. In some very real sense, there is no escaping them online. But there ought to be ways to see them less; to recommend them less; to fund them less. Other platforms have realized this as they’ve grown up. Here’s hoping Substack does the same.

Investors are focused on these metrics.

Startups should take notice.

Read the full article to learn how investors think and how you can lean into these numbers today.

Read the Article

*Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank & Trust®; Members FDIC.

Platformer has been a Mercury customer since 2020. This sponsorship gets us 5% closer to our goal of hiring a reporter in 2024.

On the podcast this week: Kevin and I discuss the Times’ lawsuit against OpenAI. Then, Beeper CEO Eric Migicovsky stops by to discuss his campaign to turn blue bubbles green on iMessage. And finally, Kevin and I trade our New Year’s tech resolutions.

Governing

Pornhub has blocked access to people living in Montana and North Carolina in protest of the states’ newly-passed age verification laws. (Wes Davis / The Verge)
Twitch is again updating its sexual content moderation policies to ban implied nudity. (Jay Peters / The Verge)
The US labor board ruled that Alphabet illegally refused to negotiate with YouTube Music contract workers who voted to unionize. (Josh Eidelson / Bloomberg)
A lawsuit against Snap brought by the family members of children who overdosed on drugs allegedly bought through Snapchat is moving forward. (Mike Masnick / Techdirt)
The New York Times copyright lawsuit against OpenAI and Microsoft hinges on the legal question of “fair use.” The outcome of the case could change the future of AI. (Will Oremus and Elahe Izadi / Washington Post)
- Related: OpenAI is reportedly offering news publishers between $1 million and $5 million annually to license news articles for training. I’ll take it! (Sahil Patel and Stephanie Palazzolo / The Information)
Gateway Pundit, a right-wing news outlet peddling false election claims, is continuing to push disinformation even as other publications pull back. (Sarah Ellison / Washington Post)
Midjourney founder David Holz, who previously insisted he dislikes fake photos, must grapple with the tool’s influence and potential for misuse as the 2024 election looms (Parmy Olson / Bloomberg)
The UK is looking to beef up its already extensive surveillance laws, sparking concerns from industry execs and privacy advocates on user privacy. (Laurie Clark / POLITICO)
The biggest misinformation threat isn’t technology or AI, but prominent politicians, the director of the Reuters Institute for the Study of Journalism argues. (Rasmus Nielsen / Financial Times)
The global cost of internet shutdowns was estimated to be over $9 billion last year, according to a report by Top10VPN. (Samuel Woodhams and Simon Migliano / Top10VPN)
Hackers are increasingly targeting verified government and business X accounts to promote crypto scams. (Bill Toulas / Bleeping Computer)

Industry

Almost half of British teenagers say they feel addicted to social media, according to data from the Millenium Cohort study. (Hannah Devlin / The Guardian)
TikTok is reportedly raising the commission it takes from Shop sellers to 8 percent, a jump from the previous 2 percent plus 30 cents per transaction fee. (Theo Wayt and Ann Gehan / The Information)
- TikTok is also reportedly aiming to grow its e-commerce business to about $17.5 billion this year, closing in on Amazon. (Zheping Huang, Alex Barinka, Dong Cao and Olivia Poh / Bloomberg)
A “GPT Store” is reportedly set to launch next week, allowing OpenAI customers to buy customized chatbots. (Stephanie Palazzolo / The Information)
Organizations on X can now get a gold check mark badge for $200 a month instead of $1,000. You can read more about this hilariously botched rollout in Zoë’s forthcoming book! (Jay Peters / The Verge)
Google is moving forward with its plan to eliminate cookies, and marketers say the ad industry is nowhere near ready. (Miles Kruppa and Patience Haggin / The Wall Street Journal)
Amazon’s crackdown on sellers has spawned a new industry – e-commerce lawyers representing sellers trying to regain access to their accounts. (Camilla Hodgson / Financial Times)
Jeff Bezos and other venture capitalists are betting on Perplexity, an AI startup that’s challenging Google’s dominance in search. (Miles Kruppa / The Wall Street Journal)
Facebook rolled out “Link History”, a tool that saves users’ browsing activity in the Facebook mobile app and uses the data to target ads. (Thomas Germain / Gizmodo)
A look at the legal battle over IRL, the social media startup that Softbank bet $150 million on and later shut down over fraud allegations. Former staffers suspect the platform was a fraud, but the founder says he was used as a scapegoat. (Joe Miller and David Keohane / Financial Times)
Before AI, there was Twitter account @Horse_ebooks managed by Jacob Bakkila, reusing and repurposing existing content as a language-based bot. (Kari Paul / The Guardian)
New PCs and laptops from Microsoft partners will have a Copilot key, the first big change to the Windows keyboard layout in almost 30 years. (Tom Warren / The Verge)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and whatever’s on your mind: casey@platformer.news and zoe@platformer.news.

OpenAI’s news blues

Casey Newton — Wed, 03 Jan 2024 01:04:59 GMT

Happy New Year! Today, let’s talk about one of the biggest pieces of news from the break — and explore why the days of freely training large language models on every public website are likely over for good.

The news, of course, is that the New York Times is suing OpenAI. Here are Michael M. Grynbaum and Ryan Mac:

The Times is the first major American media organization to sue the companies, the creators of ChatGPT and other popular A.I. platforms, over copyright issues associated with its written works. The lawsuit, filed in Federal District Court in Manhattan, contends that millions of articles published by The Times were used to train automated chatbots that now compete with the news outlet as a source of reliable information.
The suit does not include an exact monetary demand. But it says the defendants should be held responsible for “billions of dollars in statutory and actual damages” related to the “unlawful copying and use of The Times’s uniquely valuable works.” It also calls for the companies to destroy any chatbot models and training data that use copyrighted material from The Times.

First, a reminder / disclosure: I co-host the Hard Fork podcast for the Times. As such, I want to be clear any opinions here are my own, and in fact I’m not going to speak to the merits of the case at all — other than to say that I think the complaint makes for compelling reading. (Particularly the portions alleging that ChatGPT regurgitated long sections of Times articles verbatim after being prompted.)

OpenAI, for its part, told the Times reporters that it had been in negotiations with the paper and was “surprised and disappointed” by the lawsuit.

I’m interested in the case because I think generative AI has the potential to reshape the economics of journalism and the web to favor the builders of AI models over digital publishers. Already I find myself regularly looking up non-critical information via AI chatbot rather than Google, a habit that is generally faster than the alternatives but also deprives publishers of the advertising revenue they might otherwise get from me visiting their websites.

The Times case is important because it tests the legality of these fast-growing services on copyright grounds. The question is whether the specific ways that LLMs process data will be found to be covered by fair use — an as-yet unsettled but critical question for the future of digital media.

The Times reporters have the relevant argument from the complaint:

“Defendants seek to free-ride on The Times’s massive investment in its journalism,” the complaint says, accusing OpenAI and Microsoft of “using the Times’s content without payment to create products that substitute for The Times and steal audiences away from it.”

In Bloomberg, Noah Feldman weighs arguments for and against the Times’ complaint and concludes that the company is right to believe that LLMs are using its work in a way that harms its business interests. At the same time, he writes, copyright law alone will likely be insufficient to protect journalism:

Here is where the fundamental public interest in the maintenance of the free press becomes relevant to the fair use question. If you can get information more cheaply from an LLM than from the New York Times, you might drop your subscription. But if everyone did that, there would be no New York Times at all. Put another way, OpenAI and Microsoft need the New York Times and other news organizations to exist if they are to provide reliable news as part of their service. Rationally and economically, therefore, they ought to be obligated to pay for the information they are using.
Fitting this powerful public interest into copyright law won’t be simple for the courts. Literal copying is the easiest form of infringement to punish. In ordinary legal circumstances, if LLMs change words sufficiently to be summarizing rather than copying, that weakens the Times’ case. Yet summaries in different words would still be sufficient to kill the Times and similar organizations — and leave us newsless.

For these reasons, AI model makers and publishers are both incentivized to build partnerships here. And the past few months have seen OpenAI making a handful of licensing deals, including with the Associated Press and Politico and Business Insider parent company Axel Springer.

But as the Times case shows, not every publisher will come to terms with LLM makers. What then?

The recent history of Google and Meta offers one possibility.

As Google search and Facebook grew in popularity during the 2010s, they became key destinations for consumers to get news. They also built advertising products far superior to those of most publishers, and the news business continued to shrink as a result.

Platforms’ outsized success is often perceived to have come at the expense of journalism, leading to calls for Google and Meta to subsidize the industry. Most publishers continue to benefit more from Google and Meta than the reverse: tech companies send a ton of traffic to publishers, which publishers monetize via ads. But platforms can mostly live without news.

In any case, platforms did cut various deals with digital publishers over the years, just as OpenAI is seeking to do now. But as social sharing habits shifted, news became less important to Meta and Facebook than it had been in the 2010s, and the company wound down its partnerships. And while news remained more important to Google as a necessary component of search, the company’s journalisms initiatives never amounted to more than a rounding error in its annual budget.

Then in 2021, regulators in Australia had an idea. Why not tax Google and Meta for the right to display links to news publishers? The country passed a law requiring the platforms to negotiate fees with publishers or be forced into binding arbitration with an Australian panel; Meta briefly stopped displaying news links in the country as a result. In the end, though, both platforms caved.

Now similar legislation is popping up around the globe. Most recently, Canada passed a link tax of its own. New Zealand and the United Kingdom have floated their own variations.

I don’t like the idea of taxing companies for displaying links; the web relies on people being able to freely link to one another, and the laws passed in Australia and Canada are little more than shakedowns that do almost nothing to ensure that the resulting revenues are spent on journalism.

At the same time, now that two countries have shown it is possible to wring a few dollars out of platforms this way, many more seem likely to follow.

And that seems notable in the context of OpenAI, which also displays links to publishers when citing sources in ChatGPT. (Microsoft, which uses OpenAI’s APIs in its Bing search results, does so as well.)

Assuming OpenAI and its rivals continue to grow quickly — possibly at the expense of the digital publishers they are citing in links that few people ever click on — it doesn’t seem unlikely that they could be brought into a regulatory scheme similar to the one Google and Meta are now already under.

A better solution would be for AI developers to pay a fair price to any publisher from whom they are deriving significant, ongoing value. If they don’t, though, the experience of Meta and Google suggests that eventually big publishers will get paid one way or another.

Substack and Nazis

Thanks to everyone who wrote to us over the break regarding Substack’s response to criticism that it is hosting and monetizing dozens of extremist publications, including some that are openly advocating for genocide against Jewish people.

In the wake of Jonathan M. Katz’s November article on the subject in The Atlantic, 247 Substack writers published an open letter asking the company to clarify its policies. On December 21, Substack co-founder Hamish McKenzie delivered the company’s answer in a blog post. While allowing that “we don’t like Nazis either” and stating that Substack wished “no-one held those views,” McKenzie said that “some people do hold those and other extreme views.”

“Given that, we don't think that censorship (including through demonetizing publications) makes the problem go away — in fact, it makes it worse,” he wrote. “We believe that supporting individual rights and civil liberties while subjecting ideas to open discourse is the best way to strip bad ideas of their power.”

This statement appeared to be a tacit admission that Substack is not always enforcing its own content guidelines, which state that “Substack cannot be used to publish content or fund initiatives that incite violence based on protected classes.” It also seems all but certain to worsen the problem by inviting Nazis to Substack and telling them explicitly that they can make money there. (Substack takes 10 percent of subscription revenue from every paid site on the network, including Platformer.)

Content moderation often involves difficult trade-offs, but this is not one of those cases. Rolling out a welcome mat for Nazis is, to put it mildly, inconsistent with our values here at Platformer. We have shared this in private discussions with Substack and are scheduled to meet with the company later this week to advocate for change.

Meanwhile, we’re now building a database of extremist Substacks. Katz kindly agreed to share with us a full list of the extremist publications he reviewed prior to publishing his article, most of which were not named in the piece. We’re currently reviewing them to get a sense of how many accounts are active, monetized, display Nazi imagery, or use genocidal rhetoric.

We plan to share our findings both with Substack and, if necessary, its payments processor, Stripe. Stripe’s terms prohibit its service from being used by “any business or organization that a. engages in, encourages, promotes or celebrates unlawful violence or physical harm to persons or property, or b. engages in, encourages, promotes or celebrates unlawful violence toward any group based on race, religion, disability, gender, sexual orientation, national origin, or any other immutable characteristic.”

It is our hope that Substack will reverse course and remove all pro-Nazi material under its existing anti-hate policies. If it chooses not to, we will plan to leave the platform.

We’ll share our complete findings in another post when it is ready. In the meantime, we want to hear from you. Have unsubscribed or considered unsubscribing from Platformer or other Substacks over this issue? If so, email us and we’ll share a selection of the feedback with Substack.

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

X lost its bid to block a California law that requires companies to disclose content moderation policies and how they’re enforced. (Peter Blumberg and Malathi Nayak / Bloomberg)
Facebook suspended the far-right Libs of TikTok account for violating community standards, but restored it after owner Chaya Raichik appealed. (Tracy Connor / Daily Beast)
Donald Trump’s former fixer Michael D. Cohen accidentally gave his lawyer fake legal citations made up by Google Bard, not realizing that Bard is a generative AI service, he says. (Benjamin Weiser and Jonah E. Bromwich / The New York Times)
The effective altruism movement is influencing policymakers to prioritize the existential risks of AI. Now, effective altruists are clashing with AI optimists and others concerned about AI’s more immediate dangers. (Brendan Bordelon / Politico)
US Supreme Court Justice John Roberts is urging “caution and humility” as AI reshapes judicial work. (John Kruzel / Reuters)
Google will settle a 2020 lawsuit that alleged Chrome’s Incognito mode still tracks user data. (Eric Bangeman / Ars Technica)
A large network of YouTube channels spreading Chinese propaganda was leveraging AI-generated voice-overs. It got more than 120 million views before anyone at YouTube realized what was going on. Not great! (Tiffany Hsu and Steven Lee Myers / The New York Times)
Some apps that allow people to earn money, like eBay and Airbnb, will report those earnings to tax authorities, under a new international agreement that includes the US, UK, and many European countries. (Ben Lovejoy / 9to5Mac)
Microsoft president Brad Smith is walking back his criticism of the UK Competition and Markets Authority, now saying that its treatment of the Activision deal was “tough and fair”. (BBC)
EU competition chief Margrethe Vestager argues that the AI Act will enhance technology and create “legal certainty” for companies. (Javier Espinoza / Financial Times)

Industry

OpenAI reportedly topped $1.6 billion in annualized revenue due to the growth of ChatGPT — a 20 percent jump over two months. (Maria Heeter, Amir Efrati and Stephanie Palazzolo / The Information)
Alexander Reben, an MIT-educated technologist who studies the impact of innovations on creativity, will become OpenAI’s first artist-in-residence. (Leslie Katz / The New York Times)
Microsoft’s AI Image Creator, powered by DALL-E 3, can be used to generate realistic violent images of women, minorities, politicians, and celebrities. (Geoffrey A. Fowler / Washington Post)
The Microsoft Copilot app is now being rolled out on iOS and iPadOS, giving users access to GPT-4 and DALL-E 3. (Emma Roth / The Verge)
Fidelity marked down the value of its X shares again. The financial institution now says X is worth 71.5 percent less than it was when Elon Musk bought it. (Dan Primack / Axios)
X is bringing back headlines in posts that contain links, but now the font will be tiny. Whatever! (Ben Schoon / 9to5Google)
Telegram’s latest Android update has a new look for voice and video calls, and a new delete animation. (Ben Schoon / 9to5Google)
Writers could use AI the same way poets and novelists once used séances and Oujia boards – as a source for inspiration, this essay argues. (A.O. Scott / The New York Times)
Brands have found a new way to cut advertising costs – using AI-generated virtual influencers rather than real-life ones. It turns out AI influencers have fewer opinions. (Cristina Criddle / Financial Times)
People are creating AI-generated replicas of living experts without their consent.(Mohar Chatterjee / POLITICO)
Free event planning app Partiful is booming in popularity as a trendy alternative to Facebook Events. (Dan Rosenzweig-Ziff / Washington Post)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and copyrighted LLM output: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

14 predictions about 2024

Casey Newton — Fri, 15 Dec 2023 01:00:21 GMT

“The numerals 2024 set against a futuristic, utopian cityscape” / DALL-E

A year ago, as I sat down to write my predictions for 2023, two things seemed unusually clear. One was that Elon Musk’s purchase of Twitter was disrupting the old order in social networking far faster and more dramatically than even his biggest detractors had predicted. The other is that the launch of ChatGPT just a few weeks earlier had set in motion a much more significant shift — though the scope of that shift would turn out to be much wider than I guessed at the time.

Twelve months later, we’re living in a changed world. Musk killed off the Twitter brand, along with at least half of its revenues, and drove more than 100 million people onto a rival platform built by Meta. And while the shifting fortunes of our biggest social networks continue to play a critical role in shaping politics, culture, and society, the speed with which generative artificial intelligence systems are improving now seem likely to prove even more significant.

Over the next year, I expect to see both trends accelerating: on one hand, a once-in-a-generation realignment in text-based social networking that will further entrench Meta as the biggest player in the space; and on the other, the proliferation of generative AI into every digital surface that we touch, and much of the infrastructure that powers it. The trends will have their highest-profile collision to date in the run-up to the 2024 US presidential election, where platform policies and enforcement mechanisms could once again prove hugely consequential — for American democracy, and for the platforms themselves.

Just thinking about it all is enough to make me want to go lie down for a couple weeks — which, upon sending out today’s newsletter, is exactly what I intend to do. Before that, though, I have a few more thoughts on what we can expect in 2024, and Zoë has some to share as well.

In the meantime, thanks for supporting Platformer through a momentous year. If you’d like to give a subscription as a gift to someone this year, you can do so at this link.

As always, let’s start out by looking at what I told you would happen in the past year, and see how I did.

What I said: “The media will begin its divorce from Twitter. Elon Musk’s continued promotion of right-wing causes and personalities will push away more and more high-profile users, who find themselves increasingly put off by his shock-jock antics and whim-based approach to content moderation.”

The reality: This was probably the thing I got most right about 2023. The divorce began in earnest in April, when NPR left the platform after falsely being labeled “state media.” Other publishers followed, Platformer included, and found that it is entirely possible to continue growing and monetizing an audience without relying on the diminishing and artificially throttled traffic coming from Musk’s fever swamp. In July he cashed out 17 years of brand equity to re-christen Twitter as X; that same month, Meta launched Threads and rocketed to 100 million users within days.

What I said: “The use of ChatGPT in education will spark a national conversation about AI…. … By spring break, we will have seen controversies related to the use of AI in education around the country, and by year’s end I wouldn’t be surprised if OpenAI had been dragged in front of Congress to talk about it.”

The reality: All of that happened more or less exactly; OpenAI CEO Sam Altman was testifying before Congress by May. At the same time, my prediction wound up understating the impact of AI on the year’s events rather dramatically. Education is an important piece of the generative AI story, to be sure, but the technology proved just as disruptive inside law firms, marketing agencies, newsrooms, software development teams, and Hollywood, which went on strike in part due to the labor implications of studios replacing human actors with digital replicas.

And while I was right that Altman would testify, I was wrong about the subject. He did touch on the educational benefits of AI during his appearance, but most of his testimony focused on broader questions of how to regulate AI and build the technology safely.

I also did not predict that Altman would be fired by his board for unspecified reasons and be returned to his job a few days later after a near-mutiny by 95 percent of his employees.

What I said: “The Web3 vision fades into the rear view.”

The reality: Pretty much. Sam Bankman-Fried went to prison, Binance CEO Changpeng Zhao will soon join him there, and not a single crypto project managed to break through to public awareness. Crypto prices did rebound, but they have never been known to follow any particular logic.

What I said: “Content moderation will become illegal in parts of the country.”

The reality: The US Supreme Court has yet to hear the case on the social media laws in Florida and Texas; that is now set to take place in February. So I was wrong in my prediction that content moderation would become illegal in 2023 — but I might be proven right in the end. (For what it’s worth, I’m feeling significantly less confident in this prediction than I felt last year. The Supreme Court’s decision in the Taamneh case this year gave me some hope that it will respect precedent in the Florida and Texas cases as well.)

What I said: “Substack will launch an ad network.”

The reality: Wrong again. Though I still think Substack gets there in the end.

II.

And now, some predictions for next year.

Threads overtakes X in daily users and becomes the leading text-based social network. It’s hard to imagine how Threads could have had a better launch than it did. After its initial record growth tapered off, the app settled at close to 100 million monthly users. But Meta continued to push at an impressive pace; as of today, the company is both beginning to make good on its promise of linking Threads to the Fediverse and opening its doors to the European Union.

Over the next year, expect Meta to continue pushing Threads heavily in Instagram, leveraging the massive audience of its parent app to drive more daily usage of both. The arrival of an API will entice more publishers, public officials, emergency services, sports fans, and other holdouts to begin using the app more heavily. Threads won’t be feature-compete by next December, but it will be the social network that feels like home to most of the US media.

Bonus prediction: With Threads ascendant, Bluesky begins to wither as its development team prioritizes building its underlying protocol over growth, community management, and making improvements to the user experience.

Google mostly catches up to OpenAI in LLM quality and begins to neutralize GPT's lead. After it unveiled its Gemini models, Google got grief for showing what its next-generation AI systems can do in a way that overstated their capabilities. But I predict that by next December, we’ll be talking much less about the capabilities of these current-generation models — and much more about the product experiences they enable.

The reason is that by most accounts, the models are already roughly at parity. Over time, it will matter less that ChatGPT is 3 percent better at this benchmark, and Gemini Ultra is 4 percent better at that one. Instead, it’s going to come down to product and distribution. Who can build the best user experience? Who can get it in front of the most people? Microsoft and Google are best positioned to capture the upside there — but not well enough that we should expect them to inspire millions to defect from one suite of productivity apps to the other. (This was one lesson of the Bing relaunch this year.)

Google will still face plenty of challenges in 2024, but “Gemini sucks” is not going to be one of them.

GPT-5 begins training. On one hand, it hasn’t even been a year since GPT-4 was released to the public. On the other, the pace of innovation keeps accelerating — and after its leadership drama this fall, OpenAI will be eager to demonstrate that it remains the leader in AI development. Expect an announcement that work on the next frontier model has begun by the second half of the year.

The quality of search results degrades as Google proves unable to reliably detect AI-generated content. In 2024, AI-produced dreck will find its way into nearly every corner of the internet. While most of it will be inoffensive and generally correct, it will get enough wrong — and cause enough frustration among those it misleads — that trust in search will decline.

More sophisticated users will seek refuge in trusted media brands and individual creators. And Google will work to devise new signals to sort AI-written pages from human ones. But its bigger incentive is to replace search as we know it with an omniscient LLM, supported by a relatively small number of data licensing deals and an even more powerful advertising engine than the one it has today. Some will be surprised at just how willing Google is to let the web decline — but only if they choose not to look at where the puck is going.

Scaled-up AI-generated sludge outcompetes many digital media companies for advertising and affiliate-link dollars, sparking further waves of job losses and consolidation. In 2023 we lost a host of excellent publications whose work featured regularly in Platformer, including BuzzFeed News, Protocol, and Vice as we know it. And all of that happened before the rise of AI-powered content farms, which will further challenge the ability of digital publishers to sell premium ads and earn commissions on product sales. Subscriptions, live events, and podcast dollars offer a few promising paths forward, but the sheer challenge leads to more layoffs, shuttered publications, or — if the publications can manage it — an opportunistic sale.

The US presidential election is full of synthetic media, and it mostly doesn’t matter. This might barely even count as a prediction, given that Donald Trump’s troll army is already working to spread deepfakes, and getting very little attention for it. But while AI may very well play a role in the 2024 election — perhaps a hack-and-leak operation kicked off by an AI-assisted spear-phishing campaign — phony videos of the candidates seem unlikely to have much effect on the outcome. With the media fact-checking them in real time, and (most) platforms restricting their reach, it’s hard to see how synthetic media can break through in any meaningful way.

Fragmentation in social media will make it more difficult to understand which narratives are resonating most with voters. 2024 won’t be the TikTok election, or the Threads election, or the YouTube election, or the group chat election. It will be some amalgamation of all of them, along with Telegram, conservative talk radio, legacy media, and many other sources. While all of these bleed into each other, overall this looks to be the most fragmented media environment of any US national election in two decades. Following the story will take much more effort, and blind spots are likely to emerge.

Bonus prediction: Despite endless talk about wanting to eliminate it among politicians, TikTok skates unscathed through another presidential administration. After Montana’s effort to ban the app failed decisively on First Amendment grounds, enthusiasm for a similar challenge in the federal government fades rapidly.

Apple's Vision Pro is successful enough to revive interest in mixed reality and the metaverse. Despite an eye-watering average sales price of $3,700, Apple’s first major product launch since its watch will achieve its goal of selling 400,000 to 500,000 units. That’s $1.5 billion in revenue — more than enough to justify continued investment in the product — and Apple will be further buoyed by positive reviews from early adopters. While the device won’t have a ton of everyday utility, it will be good enough to justify the purchase for the kind of people for whom pricey technology serves as a status symbol. Which is to say: Apple customers.

The number of people who say they are in romantic relationships with AI companions will increase sharply. We have a national loneliness epidemic. For some people, their primary experience of most of their friends and family is already digital, with text messages and the occasional phone call making up the majority of their interactions. Into this equation add chatbots with longer memories, the permission to engage in sexually explicit dialogue, and synthesized voices. It seems obvious to me that this is a billion-dollar company just waiting to be founded — and once it succeeds, the culture war over a generation lost to chatbot romance is going to be something to behold.

Meta kills off its incomprehensible celebrity-based AI characters. In September Meta unveiled more than a dozen AI chatbots based on celebrities for people to interact with on Facebook and Instagram. While there’s plenty of reason to think people would want to chat with an AI Snoop Dogg or Tom Brady, it’s not at all clear they want to talk to avatars that merely look like them. Expect this one to die a quiet death over a holiday weekend next year.

III.

And here are two from Zoë.

Donald Trump will start ferociously posting on X once again. Donald Trump will become the Republican nominee, and once again work to galvanize his base on X. His exclusivity clause with Truth Social has expired, and the platform is “running on fumes,” according to the New York Times. Why fight the collapse when X has welcomed him back and promised a playground nearly free of content moderation? The move would be a boon for Musk, whose platform is in desperate need of revitalization, and also for Trump, who currently has 87.4 million followers on X and already rode the platform to the presidency once before. A tenuous partnership will last until Trump posts something that violates X’s rules and Musk takes it down, pitting two of the biggest egos in the United States against one another. The ensuing fight will draw attention to all involved, which is all these men seem to want anyway.

The Department of Justice will file a complaint against X on behalf of the FTC for violating its 2022 consent order. The FTC has been investigating X for more than a year to find out whether the company violated its 2022 consent order. We don’t know the results of that investigation, but we do know that the FTC has requested information about the rollout of Twitter Blue, the Twitter Files, and the layoffs that cut crucial roles in privacy and security. The fight has become highly politicized with House Republicans accusing the FTC of trying to silence Musk. If the FTC can stay its course, I predict that it finds that Musk did not conduct the detailed privacy and security reviews that are required by the consent order, nor did he work to maintain teams that could conduct privacy and security reviews. Quite the opposite, in fact! Since FTC fines tend to be compounding — and this would be the second time X would have violated its agreement — we could expect to see fines well north of $150 million.

Don’t just bank, perform at your highest level.

In order to succeed, startups need grit, talent, and the ability to unlock their full potential.

That’s where Mercury comes in. Mercury is banking and credit cards* designed to help startups operate for excellence every day.

How?

By eliminating the hurdles that come with traditional banking so you can take action.
By providing peace of mind with strategies to safeguard your funds.
By going beyond banking, and sharing access to the knowledge and network it takes to succeed.

It’s time to transform your company into the best version of itself. Leave the friction behind and join over 100K companies already on Mercury.

EXPLORE MERCURY

*Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank & Trust®; Members FDIC. The IO Card is issued by Patriot Bank, Member FDIC, pursuant to a license from Mastercard.

Platformer has been a Mercury customer since 2020. This sponsorship gets us 5% closer to our goal of hiring a reporter in 2024.

On the podcast this week: Cloudflare CEO Matthew Prince stops by to talk about how the internet changed in 2023. Plus, we talk about the future of app stores in the wake of this week’s antitrust ruling against Google, and Kevin discusses his foray into the world of the effective accelerationists.

And on the podcast next week: Kevin and I trade 2024 predictions, and then a special celebrity guest stops by to answer your hardest tech dilemmas.

Governing

UK Prime Minister Rishi Sunak is considering a possible ban on social media access for kids under the age of 16. (Thomas Seal, Kitty Donaldson and Jillian Deutsch / Bloomberg)
The UK’s Ofcom is investigating TikTok over concerns of “inaccurate” information about its parental controls. (Cristina Criddle / Financial Times)
- TikTok quietly removed rules in its terms of service that required user disputes to be handled through private arbitration. The company now requires complaints to be filed in a specific California court within a year of alleged harm. (Sapna Maheshwari / The New York Times)
- Support for a TikTok ban is waning, even among Republicans. (Drew Harwell / Washington Post)
A federal judge tentatively ruled that Elon Musk must testify again in the SEC’s probe into his takeover of Twitter. (Chris Prentice and Jody Godoy / Reuters)
Donald Trump is cheering on a team of pro-Trump creators who are pushing out racist and sexist memes ahead of the 2024 election. (Ken Bensinger / The New York Times)
Apple updated its law enforcement guidelines to raise the bar for officials seeking to obtain user push notification data. The company now requires a judge’s order. (Raphael Satter / Reuters)
Twitch is now allowing some content that shows “breasts, buttocks, or pelvic region” if the stream has a Content Classification Label on it. (Emma Roth / The Verge)
A group of Substack writers called on the company to enforce its hate speech policies and remove the Nazis openly writing and selling subscriptions on the platform. Relatedly, Today in Tabs announced today that it is decamping for Beehiiv. (Jonathan Katz / The Racket)
Ten companies – including Google, Meta and Qualcomm – are teaming up to form the Coalition for Open Digital Ecosystems, aimed at promoting more open systems in response to new EU regulations. (Foo Yun Chee / Reuters)
European regulators made a provisional deal to force rideshare and food delivery companies to classify some drivers as employees, which will include minimum wage and sick pay. (Jillian Deutsch and Aggi Cantrill / Bloomberg)
Apple is facing a potential fine and a ban on some rules in the App Store from the European Union, following allegations it used the App Store to fight off competition from music-streaming rivals. You think?? (Samuel Stolton / Bloomberg)
AI deep fakes are disrupting the elections in Bangladesh, with users paying as little as $24 a month to create AI-generated videos spreading disinformation. (Benjamin Parkin / Financial Times)

Industry

Threads is now available in the EU, home to about 448 million people. Ensuring that the phase of Threads where users simply introduce themselves will continue well into 2024. (Jon Porter / The Verge)
Meta is testing a feature to show Threads posts on Mastodon and other ActivityPub networks, Mark Zuckerberg said. Making good on a promise well before I thought the company would, even if it is only a test for now. (Ivan Mehta / TechCrunch)
Threads will adopt a fact-checking program similar to the one on Facebook and Instagram, according to Adam Mosseri. Even more good news from Threads this week. (Malek Saleh / Engadget)
ByteDance is reportedly canceling its next VR headset, the Pico 5, after sales of the last version fell short of expectations. (Juro Osawa / The Information)
OpenAI signed a licensing deal with publisher Axel Springer, in which OpenAI will pay to use content from publications like Politico and Business Insider to train ChatGPT. (Alexandra Bruell / The Wall Street Journal)
A new paper from OpenAI’s “superalignment” team describes how it is attempting to use weaker models like GPT-2 to supervise more powerful models and prevent them from doing harm. (Will Knight / WIRED)
Dropbox is doing damage control after shipping a search feature that shares data with OpenAI. The company says the data isn’t used to train AI models and is deleted within 30 days. (Benj Edwards / Ars Technica)
Chatbots like ChatGPT don’t appear to have actually increased the rate of cheating in high schools, Stanford researchers say. I’m sorry but how is this possible? (Natasha Singer / The New York Times)
Instagram is rolling out a suite of new tools so users can more easily remove spam and bots. (Malak Saleh / Engadget)
- Also, the company is rolling out a new generative AI tool that lets users edit the background of their Stories through prompts. I found it sort of clunky in my initial attempts with it. (Ivan Mehta / TechCrunch)
Information about Mark Zuckerberg’s Hawaii compound, with its 6-foot-tall walls, is shrouded by a bunch of NDAs. But the property has sparked lawsuits over land rights. (Guthrie Scrimgeour / WIRED)
Major advertisers like Disney and Comcast are increasing their ad spending on Instagram after pausing campaigns on X. (Jaspreet Singh / Reuters)
Grimes is working on a new AI-powered plush toy for kids — and it’s called Grok. Not to be confused (or is it?) with x.AI’s chatbot Grok, though the toy was apparently designed for Grimes’ son with Elon Musk, X Æ A-Xii. (Taylor Lorenz / Washington Post)
Google’s AI moves: the company is now offering its AI model Gemini Pro for enterprise customers. (Davey Alba / Bloomberg)
- Imagen 2, the second generation AI model that can create and edit images based on text, is now available to cloud customers with Vertex AI and approved access. (Kyle Wiggers / TechCrunch)
- Google is also rolling out healthcare-specific AI models to help researchers summarize doctor-patient interactions, among other things. (Ashley Capoot / CNBC)
- Google is apparently working on yet another AI assistant, this one exclusive to its Pixel devices, called Pixie. (Jon Victor / The Information)
Google will soon test a tracking protection feature that will block third-party cookies by default. (Umar Shakir / The Verge)
Amazon launched Your Books, a feature that organizes books that are bought, borrowed, or saved, in a move that competes with Goodreads, which the company also owns. (Sarah Perez / TechCrunch)
Worldcoin now supports integrations with Minecraft, Reddit, Telegram, Shopify, and Mercado Libre for its World ID. (Jacquelyn Melinek / TechCrunch)
Nandini Jammi co-founded Check My Ads after helping to deplatform far-right conspiracy theorist Alex Jones. The firm targets the revenue streams of far-right influencers by alerting advertisers and filing complaints. (Taylor Lorenz / Washington Post)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

An Epic win jolts Google

Casey Newton — Wed, 13 Dec 2023 01:00:31 GMT

“a comical scene with a large, colorful Google logo being chased by characters in a Fortnite-style environment” / DALL-E

Today let’s talk about a high-stakes ruling against Google in a closely watched antitrust case, and consider what it might mean both for the companies involved and the larger digital economy.

Late Monday, the jury deliberating in Epic Games’ lawsuit against Google ruled in favor of the Fortnite developer. It found that Google harmed Epic by creating a monopoly in in-app billing and app distribution within the Android ecosystem, illegally tying the app store and its billing system together. A series of revenue-sharing deals with developers and device manufacturers were also found to harm competition.

Epic’s lawsuit was filed in 2020. Around the same time, the company sued Apple on similar grounds, claiming Apple’s app store monopoly and ban on third-party billing systems was anticompetitive and illegal. But Epic (mostly) lost that case, which was decided by a judge.

As this new trial unfolded, Sean Hollister wrote at The Verge, “Epic v. Google turned out to be a very different case.”

He continues:

It hinged on secret revenue sharing deals between Google, smartphone makers, and big game developers, ones that Google execs internally believed were designed to keep rival app stores down. It showed that Google was running scared of Epic specifically. And it was all decided by a jury, unlike the Apple ruling.

Last week Hollister wrote a recap of the trial that, in addition to exploring the many facets of each company’s arguments, serves as a highly entertaining account of Google’s app-store related chicanery over the years: sweetheart deals to favored developers like Spotify and Netflix; successful efforts to prevent third-party app stores from launching; and the lies it told about its app store not being designed as a profit center. (The Play Store’s profit margin in 2021 was estimated during the trial to be 71 percent.)

And Epic managed to prove all that despite operating in an environment where — oops! — much of the relevant evidence was destroyed by Google, thanks to executives’ use of a default setting in Google Chat that deleted their messages automatically after 24 hours. (The judge in the case, James Donato, called this “the most serious and disturbing evidence I have ever seen in my decade on the bench with respect to a party intentionally suppressing relevant evidence,” not to mention “a frontal assault on the fair administration of justice.” He has promised to investigate.)

At Stratechery, Ben Thompson hones in on the finding that Google illegally tied its billing system to the Play Store, and suggests that this is why it mattered the case was decided by a jury rather than a judge. The jury may have been less likely to consider that Google had a justifiable business reason for tying the two together, he writes — namely, that it’s the best and most efficient way for Google to collect its compensation for building Android.

Perhaps, then, this is an argument that Google will use when it appeals — which the company plans to do, a spokesperson told me.

“The trial made clear that we compete fiercely with Apple and its App Store, as well as app stores on Android devices and gaming consoles,” said Wilson White, vice president of government affairs and public policy, in a statement shared with Platformer. “We will continue to defend the Android business model and remain deeply committed to our users, partners, and the broader Android ecosystem.”

So what does this mean for the digital economy?

Nothing, yet. In January, the parties will submit post-trial briefs and begin discussing potential remedies. Google’s will offer some idea of the grounds on which the company plans to appeal. One place you can expect the company to focus, according to a source familiar with the case: Epic’s exclusion of Apple as a competition. In antitrust cases, market definition is almost everything, and if Google can persuade an appeals court that it competes against Apple in app distribution, it might be able to get this verdict overturned.

Assume it stands, though — what then?

Epic is seeking more freedom for developers to incorporate their own internal app stores and billing systems into their apps while still making them available on the Play Store, while reducing Google’s cut of its revenues as much as possible. Google, on the other hand, will be seeking … not that.

Even though Epic did not sue for monetary damages, such an arrangement would be hugely lucrative for the company. At one point in the trial, Epic CEO Tim Sweeney — who may have been exaggerating slightly for effect — suggested that the company might make “billions” more in revenue as a result of not having to give a percentage to Google. (As you might expect, Epic hailed the ruling in a blog post.)

For Google, the consequences depend on which remedies the judge decides upon. Hollister reports that in the aftermath of Epic’s move to introduce its own billing system for Fortnite — the thing that got it kicked out of the Play Store and sparked the lawsuit — Google worried that up to 100 percent of top game developers were likely to follow suit. That would surely cut deep into the $12 billion in operating profit that Epic’s expert estimated the company earns from the Play Store.

At the same time, the markets shrugged off the ruling almost entirely: Google’s stock price dropped less than a percentage point after it was announced. Perhaps that’s because investors assume that Epic will lose on appeal, or that whatever remedy is offered ultimately gives Google a pathway to continue capturing some significant percentage of the revenues now flowing through its app store. It’s worth remembering that pushing customers toward alternative app stores and billing systems is inconvenient, and the path of least resistance for developers will always be to use Google’s own systems, even if it costs them some significant portion of revenues.

One good outcome could be for Google’s revenue share to come down enough that it enables more small- and medium-sized developers to thrive and grow their businesses. Google’s take rate has never really held back Epic in any meaningful way, but surely there are some businesses you just can’t build in a world where Google takes 30 cents of every dollar.

And that’s just as true on iOS as it is on Android. Apple may have won its case in the United States, but third-party app stores and billing systems will become legal requirements in Europe next year. They’re already a requirement in South Korea. And more such regulations are undoubtedly coming.

By now it’s clear that whatever the economic future of digital apps, Google and Apple will have to be dragged there, kicking and screaming. But it doesn’t have to be that way.

“Google and Apple both treat developers as adversaries — they try to attack our revenue streams and prevent us from competing with their products,” Sweeney told Hollister in an interview today. “They’ve built these massive self-preferencing schemes all around excluding developers and disadvantaging third-party developers. I think this is very shortsighted. I think any tech company — Apple, Google included — would be much better off in the long term if they viewed developers as awesome partners and did everything they could to support them and empower them and not get in their way financially.”

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

A federal judge allowed a class action suit against Elon Musk to proceed, ruling that his false tweets, recklessness and a poop emoji comment could constitute manipulation of stock. Sorry but the idea that Twitter stockholders are suing on the grounds that they could have gotten a better price for the company seems ludicrous.(Elura Nanos / Law & Crime)
Lina Khan’s leadership of the FTC was supposed to result in accountability from big businesses, but some are concerned about her strategic approach after a few high-profile losses. (Ankush Khardori / New York)
The US government is becoming more reliant on AI, but a top accountability watchdog says the lack of policies on non-military agency AI purchases could pose national security risks. (Brian Fung and Sean Lyngaas / CNN)
The US is expected to make about a dozen semiconductor chips funding awards next year, including some multi-billion dollar deals, US Commerce Secretary Gina Raimondo says. (David Shepardson / Reuters)
How Discord’s ineffective moderation tools and limited retention policies contributed to ongoing extremist activity on the platform, contributing to the leaking of classified information there earlier this year. (Samuel Oakford, Chris Dehghanpoor, James O’Donnell and Shane Harris / Washington Post)
Judges in the UK now have official permission to use ChatGPT to write legal rulings. Gulp! (Thomas Germain / Gizmodo)
French president Emmanuel Macron says the EU’s landmark AI Act could hamper AI development and set European tech companies behind rivals in the US, UK, and China. (Javier Espinoza and Leila Abboud / Financial Times)
Apple will reportedly allow rivals use its tap-and-go payment systems, in an apparent effort to settle European antitrust charges. (Foo Yun Chee / Reuters)
A Chinese influencer talks about the government punished him and that continues to threaten his livelihood after he used social media last year to amplify activists protesting against zero-Covid policies. (Li Yuan / The New York Times)

Industry

X is set to bring in just $2.5 billion in advertising revenue in 2023 — a roughly 50 percent drop from the year before Elon Musk took over. (Kurt Wagner / Bloomberg)
Twitter was the most powerful force in news for more than 10 years, notes this piece that looks at its impact on the news ecosystem. So far, Threads is no replacement — and maybe that’s a good thing: “Threads is to Twitter as methadone is to heroin,” says Ezra Klein. “You don’t use it for the same purpose.” (Nilay Patel / The Verge)
- Bonus treat: An archive of iconic tweets. Read them and learn. (The Verge)
Twitter’s company culture before Elon Musk was idealistic. Was it naive to think the company could be a force for good forever? (Zoë Schiffer / The Verge)
The very things that made Twitter a powerful force for social change made it a tool for rampant harassment. (Sarah Jeong / The Verge)
Sam Altman is back on stage and already defending his approach to AI development as OpenAI’s CEO, saying developers need to push some boundaries in order to realize the technology’s potential. (Saritha Rai / Bloomberg)
- Despite a company valuation of $86 billion, OpenAI’s only publicly disclosed revenue figure for the last year is $45,000 from its nonprofit parent. (Jordan Novet / CNBC)
Meta is upgrading the AI assistant on its Ray-Ban Meta smart glasses and testing multimodal capabilities that allow it answer questions based on the user’s environment. (Karissa Bell / Engadget)
YouTube removed a snippet of code that disclosed which channels and creators shared in ad and subscription revenue. (Paresh Dave / WIRED)
Google Maps released new updates that include letting users change key location settings, storing Timeline locally, and letting them delete all location data. (Allison Johnson / The Verge)
Apple is beta testing Stolen Device Protection to protect against theft by restricting certain settings when a user is away from a location their iPhone is familiar with. Good! (Joanna Stern and Nicole Nguyen / The Wall Street Journal)
Snapchat released a handful of new AI-powered features for Snapchat+ subscribers, including one that allows users to create and send AI-generated images. (Aisha Malik / TechCrunch)
WhatsApp now lets users pin messages, including emojis and stickers, in direct messages and group chats. (Ivan Mehta / TechCrunch)
Startup Essential AI raised $56.5 million from investors including from Google, AMD, and Nvidia. The company, which was co-founded by an inventor of the transformer, plans to use AI for corporate tasks like data analysis. (Sarah McBride / Bloomberg)
BeReal is rolling out two new features: Behind the Scenes, which is similar to Live Photos on iOS, and RealGroups, which allows users to share BeReals with a smaller group of friends. Hard to imagine a smaller group of friends than the number of people still using BeReal. (Amanda Silberling / TechCrunch)
The New York Times hired Quartz co-founder Zach Seward as the editorial director of AI initiatives. He’ll focus on potential uses of generative AI in the newsroom. (Alexandra Bruell / The Wall Street Journal)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and alternative billing systems: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

X restores Alex Jones

Casey Newton — Tue, 12 Dec 2023 01:00:20 GMT

(Jaap Arriens / Getty Images)

On Sunday, Elon Musk restored the notorious right-wing conspiracy theorist Alex Jones to X. The move followed a poll he posted for his followers, in which a large majority of respondents appeared to vote in favor of bringing Jones back.

Like many of Musk’s decisions, this one appeared to have arrived on a whim, after someone tagged him in a post asking him to consider the idea. If the move was shocking, it was largely because Musk had previously ruled out the idea in the name of his firstborn child, who died tragically.

By now there’s no sense working up much outrage over which right-wing extremists Musk welcomes back onto X. It has now been more than a year since X’s CEO declared a “general amnesty” for accounts that had previously been banned, resulting in the restoration of various white nationalists, QAnon affiliates, and others who had been booted from Twitter for good reason under the previous regime. At this point he could announce the returns of Cobra Commander, Skeletor, and Voldemort, and few of us would blink an eye.

But Jones’ original deplatforming five years ago stands as one of the most consequential cases of collective action among tech platforms in modern history. It’s now been long enough that some readers have asked me, in good faith, whether Jones’ case should come up for appeal.

“What’s the case for indefinite bans on social media?” one reader asked me over on Threads. “In a world where big platforms are winner-take-all markets, handing out perma bans should be the absolute last resort and only be applied with very clear guidelines.”

It’s a good question. To answer it in the Jones case, let’s refresh our memory about the events leading up to his removal from Twitter and other platforms in the late summer of 2018.

That year, the national media was nearing the peak of its interest in the platforms’ role in spreading misinformation. Jones, who advanced all manner of outlandish theories across his Infowars media empire, had effectively harnessed platform recommendation algorithms to grow his audience and revenue.

Some of those theories were harmlessly nutty, such as his famous rant warning that chemicals were being put in the water to “turn frogs gay.” But he also promoted much more dangerous theories, in particular the idea that mass shootings were staged as a pretext for the US government seizing citizens’ guns. Jones’ followers tormented families of the Sandy Hook school shooting online and off; one family moved 12 times in an effort to escape the death threats and harassment. Last year, a judge ruled that Jones must pay nearly $1.5 billion for defaming eight of the shooting victims’ families. Jones then took his company into bankruptcy, in what families called an improper effort to avoid paying.

While the harassment of Sandy Hook victims is at the center of the Jones story, it is only one aspect of his deplatforming.

Over the course of 2018, there was a steady drip of stories about Jones crossing various platform lines. That February, he received a strike on his YouTube channel for saying that survivors of the Parkland school shooting were crisis actors.

He got another strike in July, with four violations. He posted two videos that contained hate speech against Muslims, and a third that attacked transgender people, I reported at the time. A fourth violated a rule against child endangerment; it had shown an adult man pushing a child to the ground under the headline “How to prevent liberalism.”

That same month, Jones stirred another controversy during a broadcast streamed on Facebook in which he accused special counsel Robert Mueller, baselessly, of child rape. Facebook suspended him for 30 days.

As Jones’ provocations piled up, platforms faced increasing pressure to take more decisive action. In the end — quite unusually — it was Apple that acted first, removing his podcasts from its app and iTunes store on hate-speech grounds. Facebook quickly followed, removing his pages for hate speech reasons. So did YouTube, which deleted his channel.

In all cases, the reason for deplatforming Jones was not “he lied about Sandy Hook.” It was that he kept saying awful things about Muslims, transgender people, and other minorities.

After most of the platforms arrived at the decision to get rid of Jones, there was one prominent holdout: Twitter. The company offered various reasons for its inaction; in one case, then-CEO Jack Dorsey suggested that it should be up to journalists to fact-check Jones’ claims on the platform in real time. (Musk offered an echo of this when he restored Jones, saying Community Notes would check his worst impulses.) But Jones’ edgelord behavior continued, and eventually Twitter capitulated and booted him off in September.

All of which brings us back to the question: what’s the case for perma-bans on social media?

Most people who lose their accounts forever have a far shorter rap sheet than Jones. Many of them legitimately were unaware that they were breaking whatever rule got them axed, and could likely become better citizens of the platform if given a second chance. For that reason, I favor medium- and long-term suspensions over permanent bans in most cases. (Discord is doing interesting work here, as I wrote about here in October.)

Staring down the barrel of financial ruin, Jones eventually did bring himself to admit that the Sandy Hook shootings were real. He also offered a tepid apology to victims’ families, though he also quickly said he was “done apologizing.”

Meanwhile, to my knowledge, he has never apologized for the hate speech that was the real reason most platforms ultimately got rid of him. And in an appearance with Musk on Twitter Spaces yesterday, Jones implied the CIA had sponsored the legal case against him — offering a fresh reason to his remaining followers to harass the families of Sandy Hook victims, if they wanted one.

It is difficult for me to look at this set of facts and conclude that Jones is a changed man, deserving of a second chance at harnessing platforms’ recommendation algorithms to grow and monetize his following.

Musk, though, has given him that chance anyway. What does it mean for X, and the media ecosystem in general?

For X, which is in the middle of its most consequential exodus of advertisers to date, the return of Jones — and his friendly appearances on Spaces with Musk and accused rapist and human trafficker Andrew Tate — will likely only accelerate the endgame. At this point it appears that Musk would rather accept the pyrrhic victory of bankrupting his platform in the name of “free speech” than taking even the minimum steps necessary to build a viable advertising business. (And if you’re wondering what kind of advertisers are clambering to show up in between Tate and Jones in the feed, 404 Media offers a memorable example.)

For the media ecosystem overall, though, I suspect the damage will be far more contained. The threat Jones posed in 2018 came from the fact that he had access to Facebook, YouTube, Twitter, and various podcasting platforms at the same time, all of which helped him to find new audience members and revenue. They also served to amplify his ideas on Twitter in particular, which then served as a kind of collective assignment editor for much of the national media.

Had those basic dynamics remained, Jones’ return would be a cause for concern. But he still has no access to Meta’s platforms or YouTube. And while a small, shameless remnant of the press corps continues to post on X as if nothing has changed, it now rarely if ever serves as an assignment editor the way it did in the past. (One admitted exception to this was the OpenAI CEO drama, in which X served as a kind of public negotiating platform for the relevant players. Aside from that, though, I can’t tell you the last time I felt compelled to check X to follow tech news or discussions.)

Jones’ return to X will surely bring with it the usual provocations; who knows which fictional chemicals will be turning which animals gay in 2024. But on the X of today there are far fewer people available to be provoked; fewer reporters to engage with and amplify his abuse; fewer advertisers to fund the infrastructure he now hopes to rely on.

The more relevant conversations are now elsewhere, and whatever Musk may be doing on X, Jones will not be a part of them.

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

A jury ruled that Google has an illegal monopoly for its app store and billing service, offering a huge win to Epic. Penalties will be decided later; the case differed from Epic’s failed suit against Apple because it hinged on secret deals Google had struck to suppress rival app stores. (Sean Hollister / The Verge)
A US judge upheld Texas’ ban on TikTok on state-owned devices, saying it reflected reasonable data privacy concerns and rejected an argument that it stymied faculty research into the app. (David Shepardson / Reuters)
The FTC is reportedly examining Microsoft’s investment in OpenAI for potential antitrust issues. There’s no official probe yet, though. (Leah Nylen / Bloomberg)
- The UK’s Competition and Markets Authority is also gathering information from stakeholders for a potential antitrust investigation into the company. (Katharine Gemmell, Thomas Seal and Mark Bergen / Bloomberg)
- Microsoft’s response to these inquiries is that it doesn’t own a traditional stake in OpenAI, so it doesn’t control it. (Dina Bass and Leah Nylen / Bloomberg)
European lawmakers agreed to the AI Act, which includes new transparency requirements and restrictions on facial recognition. (Adam Satariano / The New York Times)
- The new law means the EU’s ahead of the US in AI regulation, though it still faces criticisms regarding exemptions for open-source models and exemptions for law enforcement to use “real-time” facial recognition. (Anthony Faiola, Cat Zakrzewski and Beatriz Ríos / Washington Post)
Microsoft is partnering with labor unions to discuss how AI will impact workers. (Jackie Davalos and Josh Eidelson / Bloomberg)
Amazon is suing REKK, an alleged international organization of thieves who have taken merchandise worth millions through a series of refund scams. (Spencer Soper / Bloomberg)
Stanford researchers say the US government should invest more in AI, just like it did during the Space Race, because the technology is too important to be left up to private tech firms. (Fei-Fei Li and John Etchemendy / The Wall Street Journal)
The White House is set to undergo a series of reviews to discuss how governments should regulate the rise of global data flows and digital trade. (Shawn Donnan and Anna Edgerton / Bloomberg)
Amazon is challenging the FTC’s case against it, saying that the agency is relying on anecdotal evidence from a handful of merchants and fails to prove that the company’s practices hurt consumers. (Spencer Soper / Bloomberg)
The US and UK are accusing the Russian government of a massive global hacking campaign that targeted elections in the UK and US energy networks and spies. (Max Colchester and Dustin Volz / The Wall Street Journal)
An organization called The Disinfo Lab claims to expose US critics of Indian prime minister Narendra Modi, but it's actually a covert operation run by an Indian intelligence officer. (Gerry Shih, Clara Ence Morse and Pranshu Verma / Washington Post)

Industry

OpenAI CEO Sam Altman’s ousting was reportedly triggered by employees alleging that he was “psychologically abusive” and pitted employees against each other in unhealthy ways. (Nitasha Tiku / Washington Post)
- The OpenAI drama exposed the cracks in the AI movement, with some alarmed by AI’s power and others excited by the massive money-making opportunity. (Tripp Mickle, Cade Metz, Mike Isaac and Karen Weise / The New York Times)
Google’s new Gemini-powered Bard is better than the old Bard, but not quite as good as ChatGPT, according to this analysis. (Kyle Orland / ArsTechnica)
TikTok is now the first mobile app that isn’t a game to generate $10 billion in consumer spending, earning a spot alongside mobile giants like Candy Crush. (Randy Nelson / data.ai)
TikTok can restart its shopping operations in Indonesia, combining TikTok Shop with GoTo Group’s e-commerce unit Tokopedia, after agreeing to invest $1.5 billion in a joint venture with GoTo Group. (Olivia Poh / Bloomberg)
Elon Musk claims Grok is supposed to be funny, but the chatbot’s humor seemingly just offers a way for Musk to be an edgelord without fully committing to the bit. (Elizabeth Lopatto / The Verge)
Meta’s 7-year-long road to end-to-end encryption for Messenger and Instagram DMs was tough, facing challenges both on the technical and political fronts. (Lily Hay Newman / WIRED)
Meta added a sneaky countdown to the Threads website, suggesting that the app will launch in Europe on Dec. 14th. (Jess Weatherbed / The Verge)
Apple is reportedly planning to make more than 50 million iPhones a year in India over the next few years, accounting for a quarter of global iPhone production. (Rajesh Roy and Yang Jie / The Wall Street Journal)
After Beeper Mini, an app that lets Android users send blue-bubble iMessages, debuted, Apple shut it down. But Beeper is working to get it back up. (David Pierce / The Verge)
Google released NotebookLM in the US, an “AI-first notebook” powered by Gemini Pro that can help with understanding and summarizing documents. (Abner Li / 9to5Google)
- You can ask NotebookLM questions about the source material on which it’s trained, and the answers will reflect both the material and the wider context of the world. It’s a writing assistant of sorts. (Steven Levy / WIRED)
Anthropic’s researchers found a new way to reduce biases in its AI model – asking it to please not take into account protected characteristics of a person or else the company will get sued. And it works! (Devin Coldewey / TechCrunch)
Text-to-image AI platform Civitai and OctoML are terminating their relationship , following an investigation that Civitai can be used to generate images that “can be categorized as child pornography”. (Emanuel Maiberg / 404 Media)
Neuroscientist Erik Hoel questions the economic viability of AI, since the industries that AI is impacting now – writing, chatting, digital art, and programming assistance – aren’t that lucrative. (Erik Hoel / The Intrinsic Perspective)
Apps and websites that use AI to undress women in photos are soaring in popularity, with an increase of more than 2,400 percent in ad links on X and Reddit, researchers say. Another huge advertising win for X. (Margi Murphy / Bloomberg)
A look at the subculture of effective accelerationism in AI, which is devoted to the progress of AI at all costs. Its supporters say it’s refreshing, but critics say it’s extreme. (Kevin Roose / The New York Times)
YouTube continues to be the top most used platform for teens, followed by TikTok, Snapchat and Instagram, a new Pew Research Center survey says. (Monica Anderson, Michelle Faverio and Jeffrey Gottfried / Pew Research Center)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and other supervillains that should come back to X: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Meta’s Oversight Board takes on the Israel-Hamas War

Casey Newton — Fri, 08 Dec 2023 01:04:49 GMT

“An image showing the scales of justice with both Facebook and Instagram branding” / GPT-4

Today let’s talk about the Oversight Board’s move to weigh in on the ongoing controversy about how Meta is moderating content related to the war between Israel and Hamas. The board is making an effort to show that it can prove useful during an unfolding crisis. But given the speed and scale of the conflict — and Meta’s cool reception to many of the board’s recent ideas — it’s unclear whether the result will go much beyond restoring a couple of posts to Facebook and Instagram.

On Thursday the Oversight Board, a Meta-funded but independent body that is empowered to make binding decisions about whether posts on the company’s apps should come down or stay up, announced it would take two cases stemming from the Israel-Hamas war. For the first time, the board said it would conduct its review on an expedited basis — meaning that its decision could come in as soon as 48 hours, and up to 30 days. (Regular decisions typically take the board about three months.)

The first case selected today concerns an Instagram post:

[It] includes a video showing what appears to be the aftermath of a strike on a yard outside Al-Shifa Hospital in Gaza City. The content, which was posted on Instagram in early November, shows people, including children, injured or dead, lying on the ground and/or crying. A caption in Arabic and English below the video states that the hospital has been targeted by the “usurping occupation,” a reference to the Israeli army, and tags human rights and news organizations. Meta initially removed the post for violating its rules on violent and graphic content.

The second is a Facebook post.

[It] shows a woman begging her kidnappers not to kill her as she is taken hostage and driven away on a motorbike. The woman is seen sitting on the back of the motorbike, reaching out and pleading for her life. The video then shows a man, who appears to be another hostage, being marched away by captors. In a caption, the user who posted the content describes the kidnappers as Hamas militants and urges people to watch the video to gain a “deeper understanding” of the horror that Israel woke up to on October 7, 2023. The user posted the content around a week after the October 7 attacks. Under its Dangerous Organizations and Individuals policy, Meta has designated Hamas as a Tier 1 dangerous organization and designated the October 7 attacks as a terrorist attack.

Meta initially said the post violated two rules: one prohibiting violence and incitement, and one prohibiting content that depicts certain terrorist incidents at the moment of the attack when victims are visible. (The latter rule was a temporary revision to Meta’s community guidelines, which will become important in a minute.)

The removal of the first post, then, speaks to anxieties that Meta is acting too aggressively in restricting speech by Palestinians who are speaking out about the terrible impact of the war on civilians. The removal of the second post reflects the opposite anxiety: that Meta is acting too aggressively to silence Israelis who are speaking out about the atrocious October 7 attacks and their aftermath.

In any case, neither removal would stand. Meta restored the first post as soon as the board told the company it was selecting it for appeal; the company restored it to Instagram behind a warning screen.

The second post is more complicated. After October 7, Meta had banned users from showing hostages being taken in the conflict. But users kept posting them anyway, often to raise awareness of the hostages’ plight, and on Tuesday — presumably after the board told Meta it was going to hear this case, too — the company revised its policy once again to allow these sorts of videos so long as they are condemning the attacks.

In both cases, then, simply by taking the case, the board prompted Meta to reverse itself. This isn’t uncommon. Over the past three years, the board’s announcement of a case coincides with Meta announcing that it has reversed its original decision even before the case can be heard. I often complain that the board hears too few cases and acts too slowly in general, but this is one way in which the board does spur quick action. If you believe that Meta erred in its initial decisions in the above cases, you should be glad that the board intervened and got them restored to Facebook and Instagram within a few weeks.

At the same time, Meta’s quick action in response to the board could have the odd effect of making the board’s expedited review moot. The Oversight Board typically only takes cases where it has strong reason to believe that Meta has made a mistake. Assuming that was the situation here, Meta has already resolved the issue. It doesn’t matter whether the decision comes in 48 hours, 30 days or 30 years — the posts have already been restored.

The board could go further, as it often does, and make policy recommendations to Meta. These are non-binding, though in practice, Meta often implements them in whole or part. (In 2021, the company partially or fully implemented 55 board recommendations, and rejected 32.)

A board spokesman wouldn’t tell me if members plan to issue policy recommendations on these cases. But we know it is concerned about over-enforcement of moderation guidelines against Palestinians generally. The board previously requested that Meta conduct a human rights assessment of its impact during an escalation in the conflict between Israel and Palestine in May 2021; the resulting report found that “Meta’s actions … appear to have had an adverse human rights impact … on the rights of Palestinian users to freedom of expression, freedom of assembly, political participation, and non-discrimination, and therefore on the ability of Palestinians to share information and insights about their experiences as they occurred.” Much of this seems to be the result of reliance on automated enforcement systems.

Assuming it does want to see some policy changes, though, the question is where the board could find leverage to meaningfully improve the system. Meta’s current policy on violent and graphic content is already fairly nuanced. While it bans gore in most contexts, the policy also carves out an exemption for the exact sort of post in the first case above:

In the context of discussions about important issues such as human rights abuses, armed conflicts or acts of terrorism, we allow graphic content (with some limitations) to help people to condemn and raise awareness about these situations.

If that’s the case, what happened? I’m speculating, but it seems likely that one of the company’s contracted content moderators — or automated systems — made a mistake. It’s a deeply unsatisfying answer, particularly given the high-stakes nature of the error. But it’s also to be expected. The board reports that the number of appeals it has received since October 7 has tripled — a sign of the surge in posts related to the conflict across Meta’s platforms.

But content moderation at Meta has not scaled with the surge. The company’s layoff-heavy “year of efficiency” included cuts to moderation teams, though the company has played down the impact of the job losses. But still, of course the company is making mistakes in moderation here. The chaotic nature of war, the stressful and even traumatic character of content moderators’ work, and the sheer flood of content are more than enough to explain why the Oversight Board was able to find two cases in which someone employed by Meta made the wrong call.

The human rights’ assessment suggests that there is something deeper here, of course: that the over-enforcement of rules against Palestinian speech is in fact a policy problem. A primary reason is that Hamas is designated as a terrorist organization, and Meta and other platforms are hyper-sensitive to accusations that they host or promote terrorist propaganda. Determining which posts in a war zone are coming directly from Hamas, and which are coming from average Palestinians, is difficult, nuanced work. Moderators being asked to make judgment calls dozens or even hundreds of times a day are bound to make mistakes — and Hamas’ status as a terrorist organization all but ensures that over-enforcement of rules against Palestinians will continue.

There is no number of cases the board could take, or speed with which it could adjudicate them, that will alter that basic dynamic. And while there are no doubt refinements to Meta’s policies that it can suggest, the far bigger issue here may be that the company lacks the ability to consistently apply its rules at scale.

That’s not a reason to give up on the whole project of the Oversight Board, which remains a valuable check on some of Meta’s worst impulses. And we should be heartened that, three years into its work, its membership has at last summoned the energy to attempt to work quickly.

But as fast as the board can work, content from the conflict in Israel is being posted even faster. As it begins to deliberate, the board should be careful not to mistake an enforcement issue for a policy problem.

On the podcast this week: Kevin and I put Google’s new Gemini Pro model through its paces. Then, we consider the prospects of Tesla’s Cybertruck. And finally, a look at some of the week’s most important AI news.

Governing

Governments are spying on Apple and Google users through push notifications, US senator Ron Wyden warns in a letter to the Justice Department. (Raphael Satter / Reuters)
The New Mexico attorney general’s office is suing Meta, alleging that Facebook and Instagram recommend sexual content to minors and promote minors’ accounts to child predators. (Katherine Blunt and Jeff Horwitz / The Wall Street Journal)
As countries around the world debate how to regulate AI, the technology is already outpacing regulators and policies. (Adam Satariano and Cecilia Kang / The New York Times)
The FTC is still trying to stop the already closed Microsoft-Activision deal, arguing that the judge who allowed it held the agency to too high of a standard. (Diane Bartz / Reuters)
The cyberattack on the US Department of Health and Human Services in March 2020, attributed to a state actor, was more serious than the department let on— and turned out to be the largest ever DDoS attack on the US government. (Jordan Robertson and Riley Griffin / Bloomberg)
The United Kingdom accused a unit of Russia’s Federal Security Service of being behind a campaign to hack and leak information about the country’s government officials. (Alexander Martin / The Record)
European regulators are reportedly set to deem iMessage not an “important gateway” for business users, exempting it from regulation under the Digital Markets Act. So green text bubbles it is. (Samuel Stolton / Bloomberg)
AI could revolutionize mass spying and surveillance, security technologist Bruce Schneier warns, and corporations and governments could weaponize that. (Schneier on Security)

Industry

TikTok is reportedly planning a slew of initiatives to combat hate, including a social media campaign featuring creators called “Swipe Out Hate”, following complaints of antisemitism and other hate speech on the platform. (Kaya Yurieff / The Information)
- TikTok’s most viral videos this year include a creator’s makeup look, a DIY Iron Man arc reactor ceiling light, and a very large kitten. (Mia Sato / The Verge)
ByteDance is reportedly offering to buy back up to $5 billion worth of shares from investors as its IPO plans stall. (Zhou Xin / South China Morning Post)
A whole bunch of Meta updates today:
- Messenger is getting end-to-end encryption by default for direct messages and calls. A big deal and a good thing. (Jay Peters / The Verge)
- Tags – like hashtags, but with more words and special characters – arrived globally on Threads. (Jay Peters / The Verge)
- Meta’s celebrity-based AI characters are now live in the US for people to chat with on WhatsApp, Messenger and Instagram with support for Bing. Some characters will also “remember” conversations. (Sarah Perez / TechCrunch)
- Imagine with Meta, the company’s standalone AI image generator powered by its Emu model, is now available. (Kyle Wiggers / TechCrunch)
- Meta AI will now let group chat users create AI images using prompts that can build on top of each other with a new feature called “reimagine.” (Sarah Perez / TechCrunch)
- WhatsApp voice messages can now be set to disappear after being listened to once. (WhatsApp)
- Many Instagram videos posted before late 2014 have lost their audio, an issue that seemingly started this year. Meta said it’s working to fix the issue. (Jay Peters / The Verge)
Elon Musk ranted that Bob Iger should be fired after Disney pulled its advertising from X. (Oliver Darcy / CNN)
x.AI’s snarky “anti-woke” chatbot, Grok, is now being rolled out to X Premium Plus users. (Maxwell Zeff / Gizmodo)
The AI-powered “Help me write” feature is coming to Chrome, which will rely on both user prompts and context on the site users are writing on. (Kyle Bradshaw / 9to5Google)
Early reactions to Google’s Gemini Pro haven't been great – the model gets basic facts wrong, makes mistakes in translation, and, in response to questions about controversial news, says to Google it. (Kyle Wiggers / TechCrunch)
Amazon seems to be struggling to catch up with generative AI, and inaccuracies in its Q chatbot aren’t helping. (Corey Quinn / Last Week in AWS)
OpenAI’s Sam Altman was named TIME’s CEO of the Year, one year after the release of ChatGPT and following a tumultuous Thanksgiving for Altman. (Naina Bajekal and Billy Perrigo / TIME)
- Helen Toner, one of four OpenAI board members who fired Altman, said the move wasn’t about safety — it was about a lack of trust in ensuring that AI systems are built responsibly. (Meghan Bobrowsky and Deepa Seetharaman / The Wall Street Journal)
Twitch is planning to shut down its operations in South Korea, with CEO Dan Clancy saying the country is too expensive to operate in, despite efforts to reduce costs. (Jordan Fragen / VentureBeat)
The new Mammoth 2 app promises to make Mastodon much simpler for users, including features like Smart Lists, similar to the old lists on Twitter. (David Pierce / The Verge)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and policy recommendations: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Google unveils Gemini

Casey Newton — Wed, 06 Dec 2023 15:01:07 GMT

(Google)

Google this morning announced the rollout of Gemini, its largest and most capable large language model to date. Starting today, the company’s Bard chatbot will be powered by a version of Gemini, and will be available in English in more than 170 countries and territories. Developers and enterprise customers will get access to Gemini via API next week, with a more advanced version set to become available next year.

How good is Gemini? Google says the performance of its most capable model “exceeds current state-of-the-art results on 30 of the 32 widely-used academic benchmarks used in LLM research and development.” Gemini also scored 90.0% on a test known as “Massive Multitask Language Understanding,” or MMLU, which assesses capabilities across 57 subjects including math, physics, history and medicine. It is the first LLM to perform better than human experts on the test, Google said.

Gemini also appears to be a very good software engineer. Last year, using an older language model, DeepMind introduced an AI system named AlphaCode that outperformed 54 percent of human coders in coding competitions. Using Gemini, Google built a next-generation version named AlphaCode 2. The sequel outperformed an estimated 85 percent of humans, the company said.

Competitive coding is meaningfully different from day-to-day software engineering in some important ways: it can be both more and less difficult than what the typical engineer is asked to do. But still, the rate of progress here is striking.

Gemini is natively multimodal, meaning that it can analyze the contents of a picture and answer questions about it, or create an image out of a text prompt. During a briefing on Tuesday, a Google executive uploaded a photo of some math homework in which the student had shown their calculations leading up to the final answer. Gemini was able to identify at which step in the student’s process they had gone awry, and explained their mistake and how to answer the question correctly.

“Multimodal” can read like awkward jargon, but the term comes up constantly in conversation with Google executives. The ability of AI systems to take different kinds of data (text, images, video, audio), analyze them using a single tool, and translate them in and out of various formats is the kind of foundational innovation that makes lots of other progress possible. (All of which is a long way of saying: sorry for the number of times the word “multimodal” appears in the interview below.)

In accordance with Google’s preference for chaotic branding, Gemini will be available in three “sizes”: Nano, which is small enough to fit in a smartphone and will power features in the Pixel 8 Pro smartphone starting today; Pro, which now powers Bard; and Ultra, which will begin making its way into products next year.

Without having used any of these models, it’s difficult to compare them to those from rivals like OpenAI and Anthropic. But my basic sense is that Gemini Pro is best seen as the company’s answer to OpenAI’s GPT-3.5: In its announcement blog post, the company noted that Pro performed better than GPT-3.5 in many but not all benchmarking challenges.

That sets up Ultra as the top rival to GPT-4.5 Turbo for the crown of best widely available, general-purpose LLM. And Ultra won’t be available next year so that Google can complete trust and safety testing.

When it is available to consumers, Ultra will power a new chatbot that the company is calling Bard Advanced. While the company would not confirm it on Tuesday, the branding suggests to me that Advanced could be Google’s answer to ChatGPT Plus: a subscription-based paid product for users who want the best product available.

From there, Google says, Gemini will begin to permeate throughout the company’s ecosystem of consumer and enterprise products, starting with search, Chrome, ads, and its productivity apps.

II.

A few hours after getting briefed on the news, I had a chance to meet virtually with Google CEO Sundar Pichai and Google DeepMind co-founder and CEO Demis Hassabis.

It was my first chance to talk about the state of the art in AI with Pichai since March, when he came on Hard Fork; and my first ever conversation with Hassabis. Over 30 very fast minutes, we talked about Gemini’s novel capabilities, how AI is changing search, and whether Pichai thinks he’ll hire fewer software engineers next year as a result of the company’s progress.

Highlights of the conversation follow; this interview has been edited for clarity and length.

Casey Newton: Today you shared a variety of industry benchmarks that show the progress you’ve made with Gemini. But I’m curious about your own, personal testing of the models. What are you noticing about them that makes you feel like you’ve taken a step forward?

Demis Hassabis: I think you'll see this just by using the new Bard — the quality overall is just massively improved over our previous models. The kind of thing I’m interested in specifically is using it as a science assistant. Actually parsing scientific papers, the graphs in those papers, interpreting them. Putting tables into graphs, extending graphs. It’s been super useful, and I want to double down on that.

Sundar Pichai: Multimodality is very exciting. We’re working to get it connected to the products and expose it thoughtfully, but I think that’s where a lot of the new synapses will come in.

To me what’s exciting is that this is just our 1.0. There’s such a strong roadmap of innovation as we look into 2024. And one of the things that Demis and his team are really good at is the relentless way of iterating and coming up with new versions.

Earlier today I asked Eli Collins, the vice president of product at DeepMind, whether Gemini had demonstrated any novel capabilities. He basically told me, "stay tuned." Do you believe this model will turn out to have capabilities beyond that of previous LLMs, or do you see it as more evolutionary?

Hassabis: I think we will see some new capabilities. This is part of what the Ultra testing is for. We’re kind of in beta — to safety-check it, responsibility-check it, but also to see how else it can be fine-tuned.

Your blog post describes how much better Gemini is at reasoning. If that’s the case, I wonder how good it might be at planning. Can you envision building agents using the Gemini, for stuff like making reservations?

Hassabis: You hit the nail on the head there, Casey. That’s something we’re thinking very heavily about. It’s in our heritage, really, back from the old DeepMind Days. We’re experts at these kinds of agent-based systems and planning systems. So watch this space. We’re pushing hard on that.

But multimodality is an important thing — it’s a basic thing you need (to build agents.) If you imagine robotics, or even just digital agents, and understanding user interfaces and how to interact with things. You have to parse the environment you’re in, multimodally, before you can act in the world in a useful manner. So you can think of it as a prerequisite to the planning and interacting stage.

Pichai: But those are innovations to come.

You’re now saying that Gemini will come to search next year. How do you see it changing the search experience?

Pichai: We’re already experimenting with it in the search generative experience, and as we are experimenting with it, it’s driving improvements across the board. We think about Gemini as foundational — it will work across all our products. Search is no different.

One of the things search has been pushing hard on is multimodality in general. But today they have had to do all the hard work of making search multimodal. Gemini as a foundational model gives them that capability [natively], so I think that’s an area where they will innovate.

Do you think over the medium term, Gemini in search increases the number of times that you get the information you need from the results page without having to visit a website?

Pichai: Our fundamental vision is that people come to search to experience the richness and the diversity of the web and the content ecosystem. So even though with search generative experience we can expand what we do, we’re actually designing the product in a. way so that people can go explore. And I think that’s what users want. I view it as a fundamental value proposition of search, so that’d be part of our goal as we evolve the product.

I’m also reading that Gemini is coming to Chrome. What can you do with Gemini in a web browser?

Pichai: It can look at what’s on the web page and answer questions for you, and help you with tasks related to that. You can imagine looking at something you want to understand, like a set of figures on a web page, and saying “quickly summarize this for me.” All that is now possible, right? Again, it’s this notion of being an assistant for the user to help them with what they’re doing while they’re browsing the web. All these are possibilities.

I want to get a sense of the state of the art. I imagine that you can spend most of 2024 just refining Gemini 1.0. But as you start to look forward to training a Gemini 2.0, does that feel like a matter of simply throwing more data and compute at techniques you’ve already developed? Or are there some fundamental research breakthroughs that you would need to make first?

Hassabis: Great question. I think that the answer is both — we’re going to push the frontiers of both. We’re looking at lots of blue-sky research on things like planning; lengthening context windows; and all of these critical capabilities that current systems don’t have and that we’re going to need if we’re going to get towards AGI-level systems. So we’re hard at work on all of that.

There’s a lot more juice left to get from scaling, improving architectures, and maybe more incremental improvements, on top of these big innovative new capabilities. And there’s actually an enormous number of research areas that are looking promising.

Pichai: I would say that it feels very early to me. We have a clear line of sight that Gemini 2.0 is going to be much better. If I look at all the work that Google DeepMind is doing, and you say that there are 10 to 15 areas — right now you’re seeing fast progress in one area, right? But there are going to be innovations from the other areas, too, which will come into all of this.

Your model seems to be getting really good at winning coding competitions. A year from now, can you imagine them being good enough that you don’t need to hire as many engineers?

Pichai: I really think of this as making programmers far more productive, and over time taking some of the grunt work out of the job.

I think programmers will have such sophisticated tools that more people will be able to become programmers. We shouldn’t underestimate that. The bar will change, and it will expand access the field.

Sundar, earlier this year we were talking, and you mentioned that you wouldn’t mind if the pace of development in the AI field slowed down a bit. How are you feeling about the pace of development now?

Pichai: I have two lenses which I use. I’m very optimistic about the potential. For example, if I take a step back and say that the breakthroughs here may help us make progress against cancer more easily, I want it to move as fast as possible. Why wouldn’t you? But I do think as we are driving toward more and more capable models, we need to take time to make sure we put safeguards in place.

I think the pace right now is at a place where it’s exciting. But there will be moments where we feel like we can all take a breather and catch up, too. I think they’ll go hand in hand.

Hassabis: I agree. It’s been a bit of a rocket-ship journey for the whole field. I’ve been working on this for 20, 30 years, and for me to see this all working is fantastic. Diseases really will be cured by AI-enabled technologies. New materials will help us with climate change. The number of things that I think AI can be applied to to help society is almost limitless. We’re really close now to actual, practical, useful things in the real world, beyond just the games and stuff we used to do so well.

But at the same time, I’ve always believed it’s going to be one of the most transformative technologies that humanity will ever invent. I think more people are coming around to that view now. So we really need to be thoughtful and responsible, and have as much foresight as possible about the unintended consequences.

Investors are focused on these metrics.

Startups should take notice.

Read the full article to learn how investors think and how you can lean into these numbers today.

Read the Article

*Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank & Trust®; Members FDIC.

Platformer has been a Mercury customer since 2020. This sponsorship gets us 5% closer to our goal of hiring a reporter in 2024.

Governing

Meta, IBM, and more than 40 other companies are forming an alliance focused on responsible open-source development. The group plans to share technology, including safety tools. (Alex Barinka / Bloomberg)
The UK’s Ofcom is publishing a new set of rules for how to age-gate porn online, under the Online Safety Act, including external verification methods like credit cards and mobile carriers. (Jon Porter / The Verge)
Security technologist Bruce Schneier discusses how AI systems will increasingly confuse two types of trust — interpersonal and social — and how regulating organizations that control and use AI is essential. (Schneier on Security)
X was granted money transmitter licenses from three additional states, bringing the total number to 12. If you transmit money using X let us know how it goes for you! (Sarah Perez / TechCrunch)
AI model Civitai, backed by venture capital firm Andreesen Horowitz, reportedly generated content that “could be categorized as child pornography”. Another huge win for the accelerationists. (Emanuel Maiberg / 404Media)

Industry

Threads is now outpacing X in new downloads amid an exodus of major advertisers from X. (Sarah Perez / TechCrunch)
- But publishers are still hesitant to invest resources into Threads despite some organizations seeing growth in engagement, citing limited data available to their audience teams. (Sara Guaglione / Digiday)
The launch of ChatGPT a year ago triggered the AI arms race in Silicon Valley, as this detailed piece recounts. (Karen Weise, Cade Metz, Nico Grant and Mike Isaac / The New York Times)
Microsoft Copilot will get support for GPT-4 Turbo soon, along with other new OpenAI models, a new code interpreter, and deep search for Bing. (Tom Warren / The Verge)
Gmail’s new AI-powered spam filters says it can now understand “adversarial text manipulations” – junk emails with many special characters that went previously undetected. It should loan the technology to the Meta team responsible for catching new CSAM-related hashtags. (Ron Amadeo / Ars Technica)
Musk’s x.AI filed to raise up to $1 billion in an equity offering, a month after the debut of its chatbot Grok. (Hayden Field / CNBC)
Beeper Mini is a new Android app that reverse-engineered iMessage to allow Android users to send blue-bubble texts to their friends on iOS. Will Apple let it stand? (Jacob Kastrenakes / The Verge)
Discord’s mobile app got a redesign, with a new set of navigation tabs named Servers, Messages, Notifications, and You. (Sarah Fielding / Engadget)
Alibaba’s new “Animate Anyone” AI feature is actually trained on scraped videos of famous TikTokers dancing. (Samantha Cole / 404Media)

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and Gemini benchmarks: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

How Facebook helps predators find each other

Casey Newton — Tue, 05 Dec 2023 01:01:03 GMT

Programming note: To accommodate some news, the next edition of Platformer will arrive Wednesday morning, instead of at 5PM PT Tuesday as usual.

(Omar Marques / Getty Images)

In 2019, YouTube realized that it had a problem. Parents who had uploaded seemingly innocuous footage of their children playing in swimsuits were surprised to find that some of the videos were getting hundreds of thousands of views.

It turned out that the company’s recommendation algorithms had unwittingly created a catalog of videos of young children in various states of undress and were serving them up to an audience of pedophiles. “YouTube never set out to serve users with sexual interests in children,” Max Fisher and Amanda Taub wrote in the New York Times, “but in the end… its automated system managed to keep them watching with recommendations that he called “disturbingly on point.”

I thought about YouTube’s predatory playlists over the weekend while reading how Meta’s systems have been discovered to operate in a similar way. In a new investigation, the Wall Street Journal examined how automated systems in Facebook and Instagram continuously recommend content related to pedophilia and child sexual abuse material.

Here are Jeff Horwitz and Katherine Blunt:

The company has taken down hashtags related to pedophilia, but its systems sometimes recommend new ones with minor variations. Even when Meta is alerted to problem accounts and user groups, it has been spotty in removing them.
During the past five months, for Journal test accounts that viewed public Facebook groups containing disturbing discussions about children, Facebook’s algorithms recommended other groups with names such as “Little Girls,” “Beautiful Boys” and “Young Teens Only.” Users in those groups discuss children in a sexual context, post links to content purported to be about abuse and organize private chats, often via Meta’s own Messenger and WhatsApp platforms.

The Journal’s report follows an earlier investigation in June that documented how Instagram is used to connect buyers and sellers of CSAM. That report found that viewing even a single account in a criminal network was enough to get “suggested for you” recommendations for buyers and sellers on the service, and that “following just a handful of these recommendations was enough to flood a test account with content that sexualizes children.”

A follow-up report in September by the Stanford Internet Observatory, which aided in the Journal’s investigations, found that Meta had made some progress — but that obvious gaps in enforcement remain.

Let’s stipulate that CSAM is an industry-wide issue, and that Meta has invested more in building child-safety features than many other social networks. This year Stanford found equally disturbing CSAM issues at networks including X, Telegram, Mastodon, and decentralized networks generally. That Meta is subject to more scrutiny on this subject is less of a commentary on the company’s unique negligence than on its vast scale and the responsibilities that come with it.

Let’s also stipulate that this is what tech policy folks call an adversarial problem: a cat-and-mouse game in which predators change tactics continuously to evade every new enforcement tactic the platforms roll out. Even when platforms manage to close off some vector for harm, motivated actors will work diligently to find new ones. It’s exhausting, essential work.

Even if you’re willing to grant Meta all that, though, the Journal’s report makes for some deeply worrisome reading.

I’ve written in the past about the difference between internet problems and platform problems: “Internet problems arise from the existence of a free and open network that connects most of the world; platform problems arise from features native to the platform.”

Internet problems reflect the world we live in and can’t be blamed on any one actor. Platform problems, though, are the responsibility of the people who design and oversee them.

Viewed through this lens, the fact that pedophiles create accounts on social networks and attempt to organize there is an internet problem. The fact that algorithmic recommendations work to connect these people and create a market for CSAM and other harms is a platform problem — Meta’s platform problem.

The Journal documents several instances of surprisingly large Facebook groups that appear to be devoted to promoting problematic content — and are themselves promoted by the company’s “groups you should join” recommendations. “In one public group celebrating incest,” the authors write, “200,000 users discussed topics such as whether a man’s niece was ‘ready’ at the age of 9, and they arranged to swap purported sex content featuring their own children. In another user group numbering 800,000, administrators shared images of schoolgirls as a way to promote a Spanish-language website with a name referring to women’s underwear.”

In another case, the reporters write, “Facebook’s ‘groups you should join’ feature has suggested topics such as kidnapping, ‘dating’ children as young as 11 and even chloroforming women.”

Meta’s enforcement systems have also been easily defeated by predators who lightly modify problematic hashtags after the company bans more straightforward versions. And in other cases — incredibly — the company’s recommendation systems appear to be working directly on behalf of the pedophiles:

On a Journal Instagram test account, Meta wouldn’t allow search results for the phrase “Child Links,” but the system suggested an alternative: “Child Pornography Links.” After Meta blocked that term following a query by the Journal, the system began recommending new phrases such as “child lingerie” and “childp links.”

The design problems are so pervasive — and the violating content so easily found by outsiders — that it is difficult to believe that the teams at Meta who are charged with policing this material are adequately staffed. Thousands of layoffs in the company’s “year of efficiency” did not spare content moderation teams. The Journal reports that “hundreds” of safety staffers have been cut, though Meta says most did not work primarily on child safety issues.

The first and most important reason to address these issues, of course, is to protect the victims and the children using Meta’s platforms. But there are also compelling business reasons for Meta to be doing a better job here — and that’s the main reason I find myself surprised that the company isn’t doing a better job here.

In March, Utah became the first state to ban the use of social media apps without parental permission for children under the age of 18. Arkansas followed with a similar law, though its implementation was blocked by a judge. Meanwhile, with the release of the surgeon general’s warning about kids and social media this spring, it has now become bipartisan conventional wisdom that social media is not safe for children.

Against this backdrop, the last thing Meta needs is a series of regular, detailed reports about its apparently unmanageable child predator problem.

For its part, Meta responded with a 1,300-word blog post about all the steps it has taken since June, including setting up a child safety task force with as many as 100 employees and participating in Lantern, a coalition of tech companies that shares signals about predators between platforms so that they can be more quickly removed.

The company says it has removed hundreds of thousands of accounts that violated its child safety policies, along with thousands of groups and dozens of networks dedicated to abuse.

“Child exploitation is a horrific crime and online predators are determined criminals,” Meta told me in a statement. “We work hard to stay ahead. That’s why we hire specialists dedicated to online child safety, develop new technology that roots out predators, and we share what we learn with other companies and law enforcement. We are actively continuing to implement changes identified by the task force we set up earlier this year.”

But after reading the Journal and Stanford reports, it’s worth asking whether those task force-approved changes will be enough to address the problem. These stories offer more than the usual lists of isolated incidents — they reflect ongoing, systemic problems created in part by the company’s own machinery.

At the very least, Meta should be re-thinking how it recommends groups and hashtags. It has already taken steps to prevent what it calls “potentially suspicious adults” from being able to see each others’ comments or be suggested to one another as follows.

The truly tragic thing, given YouTube’s highly publicized experience above, is that Meta could have taken these steps at any time in the past few years. The platform dynamics of this terrible abuse are very well known — the only question has been when the company would at long last get around to addressing them.

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

Epic v. Google: Judge James Donato said he would pursue an independent investigation into Google for intentionally destroying evidence by automatically deleting internal chat messages. (Sean Hollister / The Verge)
Google reached a $27 million settlement in a lawsuit by employees alleging unfair labor practices. (Reed Albergotti / Semafor)
Google is asking the UK’s Competition and Markets Authority to take action against Microsoft for its cloud computing dominance, saying that its business practices put competitors at a disadvantage. I wonder what it would say about the market for web search? (Martin Coulter / Reuters)
Thousands of fake social media accounts were created by someone in China in an attempt to spread polarizing US political content ahead of the elections. (David Klepper / Associated Press)
Meta is facing a $600 million lawsuit by a group of 83 Spanish media companies, citing unfair competition in the advertising market. (Inti Landauro / Reuters)
Meta received another request for information about child safety on Instagram under the European Digital Services Act, including its rules around self-generated child sexual abuse material. (Natasha Lomas / TechCrunch)
Misinformation researcher Joan Donovan is accusing Harvard of ending her research project after complaints by former Meta executives linked to the school. (Joseph Menn / Washington Post)
Generative AI is reportedly causing conflict for European lawmakers discussing the AI Act, with disagreements on how systems like ChatGPT should be regulated. (Supantha Mukherjee, Foo Yun Chee and Martin Coulter / Reuters)
French prime minister Élisabeth Borne signed a directive for all government employees to uninstall foreign messaging apps like Signal, WhatsApp and Telegram in favor of French app Olvid. (Bill Toulas / Bleeping Computer)
Albuquerque Public Schools employ filters that are misguided and inappropriate, blocking students from accessing online resources like suicide prevention, race, ethnicity and LGBTQ topics, an investigation found. (Todd Feathers and Dhruv Mehrota / WIRED)

Industry

Spotify is cutting about 1,500 jobs, about 17 percent of its workforce, in its third round of layoffs this year, citing slow economic growth and rising capital costs. Also it’s canceling the wonderful podcast Heavyweight, which is insane. (Manish Singh / TechCrunch)
ByteDance is reportedly working on a platform that will allow users to create their own AI chatbots — it will be launched in beta by the end of the month. (Coco Feng / South China Morning Post)
OpenAI reportedly agreed to invest $51 million in AI chips from startup Rain AI, in which CEO Sam Altman has personally invested more than $1 million. (Paresh Dave / WIRED)
- Microsoft’s partnership with OpenAI propelled it ahead in the AI race. But while the company had ambitious AI plans to balance safety with innovation, its cautious approach turned upside down with CEO Sam Altman’s firing. (Charles Duhigg / The New Yorker)
OpenAI is delaying the launch of its announced ChatGPT store until early next year. (Ina Fried / Axios)
Asking ChatGPT to repeat words forever, which can be used to get the model to leak its training data, is now considered a terms of service violation. (Jason Koebler / 404Media)
Brad Lightcap, OpenAI’s COO, discusses the unexpected success of ChatGPT, how Altman thinks, and where he expects the technology to advance over the next year. (Hayden Field / CNBC)
X is now turning towards small and medium-sized businesses for advertising revenue, following an exodus of big advertisers after Elon Musk’s antisemitic post. Good luck with that!! (Hannah Murphy and Daniel Thomas / Financial Times)
Meta executives say there are no commercial downsides to openly sharing its AI technology, as the models could potentially improve faster with more developers working on it. (Aisha Counts / Bloomberg)
Meta AI researchers announced Seamless Communication, a new suit of AI models developed to bridge language barriers and enable more natural and authentic communication. (Michael Nuñez / VentureBeat)
Threads won’t be getting chronological search results anytime soon, as Instagram chief Adam Mosseri says they would create a “substantial safety loophole”. (Jay Peters / The Verge)
WhatsApp’s new update introduces a feature that allows users to preserve photo and video quality by sending the attachments as documents. (WABetaInfo)
Mark Zuckerberg sold Meta shares for the first time in two years, earning about $185 million after a 172 percent surge in the company’s stock price. (Benjamin Stupples / Bloomberg)
Meta is starting to disconnect Facebook and Instagram messenger chats. I assume this is related to encryption? (Kyle Bradshaw / 9to5Google)
AI-powered stars using the likeness and voices of celebrities are on the rise on platforms like Facebook, Instagram, and YouTube, despite the complexities around licensing and intellectual property. (Alex Weprin / The Hollywood Reporter)
YouTube is cracking down on ad blockers, but ad blockers are developing new strategies to counter the rules. (Anthony Ha / Engadget)
Telegram added new features to channels, including improved discovery of other channels, emoji customizations for reactions, and stats for stories. (Ivan Mehta / TechCrunch)
Bluesky is rolling out automated tools for content moderation designed to flag content that violates its community guidelines. (Sarah Perez / TechCrunch)
Kick, a live streaming platform that emerged as a Twitch competitor, cut lucrative deals with top gaming creators to lure their audiences onto the site. But criticisms of its status as an online casino and lack of content moderation remain. (Kellen Browning / The New York Times)
Business Insider promoted an article on a subreddit known for harassment, highlighting a dilemma lots of news outlets are facing — how to drive traffic as X continues to decline. (Taylor Lorenz / Washington Post)
Guillaume Verdon, a former Google engineer and founder of AI startup Extropic, is behind the @BasedBeffJezos account on X, which promotes “effective accelerationism” — advancing technology and capitalism at all costs. (Emily Baker-White / Forbes)

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and potentially suspicious adults: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Amazon’s Q has ‘severe hallucinations’ and leaks confidential data in public preview, employees warn

Zoë Schiffer — Sat, 02 Dec 2023 00:58:37 GMT

Three days after Amazon announced its AI chatbot Q, some employees are sounding alarms about accuracy and privacy issues. Q is “experiencing severe hallucinations and leaking confidential data,” including the location of AWS data centers, internal discount programs, and unreleased features, according to leaked documents obtained by Platformer.

An employee marked the incident as “sev 2,” meaning an incident bad enough to warrant paging engineers at night and make them work through the weekend to fix it.

Q’s early woes come at a time when Amazon is working to fight the perception that Microsoft, Google, and other tech companies have surpassed it in the race to build tools and infrastructure that take advantage of generative artificial intelligence. In September, the company announced it would invest up to $4 billion in AI startup Anthropic. On Tuesday, at its annual Amazon Web Services developer conference, it announced Q — arguably the highest-profile release in the series of new AI initiatives the company unveiled this week.

In a statement, Amazon played down the significance of the employee discussions.

“Some employees are sharing feedback through internal channels and ticketing systems, which is standard practice at Amazon,” a spokesperson said. “No security issue was identified as a result of that feedback. We appreciate all of the feedback we’ve already received and will continue to tune Q as it transitions from being a product in preview to being generally available.”

After publication, the spokesperson sent another statement pushing back on employees’ claims: “Amazon Q has not leaked confidential information.”

Q, which is now available in a free preview, was presented as a kind of enterprise-software version of ChatGPT. Initially, it will be able to answer developers’ questions about AWS, edit source code, and cite sources, Amazon executives said onstage this week. It will compete with similar tools from Microsoft and Google but be priced lower than rivals’, at least to start.

In unveiling Q, executives promoted it as more secure than consumer-grade tools like ChatGPT.

Adam Selipsky, CEO of Amazon Web Services, told the New York Times that companies “had banned these A.I. assistants from the enterprise because of the security and privacy concerns.” In response, the Times reported, “Amazon built Q to be more secure and private than a consumer chatbot.”

An internal document about Q’s hallucinations and wrong answers notes that “Amazon Q can hallucinate and return harmful or inappropriate responses. For example, Amazon Q might return out of date security information that could put customer accounts at risk.” The risks outlined in the document are typical of large language models, all of which return incorrect or inappropriate responses at least some of the time.

But more acute issues began to surface on Slack this morning after a program manager posted a question in Slack. “Hello everyone. Seeing any misleading answers from Amazon Q on digital sovereignty? Please contact me, I’m trying to collect before reaching out to the Service team. Thanks!”

Digital sovereignty refers to storing data in the proper legal jurisdiction — an area that has been increasingly heavily regulated in recent years, particularly in Europe. The Slack message suggested that Q might be giving users bad legal advice on the subject.

“This is more than a DigiSov topic issue,” a colleague responded, before posting the security ticket regarding the hallucinations and leaked confidential information.

Amazon tracks security incidents on a homegrown software system called SIM, which creates tickets for bugs and other problems in a fashion similar to the better-known Jira.

“...and that ticket has been locked down,” another employee wrote. “Expect the Q team to be very, very busy for a while. I’ve also seen apparent Q hallucinations I’d expect to potentially induce cardiac incidents in Legal.”

An AWS manager noted “it goes without saying that this should not be discussed in a public channel.” And yet employees continued to discuss the issue in Slack. One noted: “I expect the [correction of error] report to be a real page-turner, if we’re ever allowed to see it.” “Correction of error” is how Amazon refers to postmortem reports, in which engineers work to describe the root causes of technical problems and identify fixes in the aftermath of incidents.

The program manager who started off the conversation asked if he should open a specific ticket to discuss the problems with digital sovereignty. “I wouldn’t suggest it,” a colleague said. “The broad and egregious nature of Q’s hallucinations yesterday (including potential Legal exposure) has been well reported in the existing ticket.”

Update December 1, 11:22PM ET: Added a statement from Amazon spokesperson sent after publication.

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and severe Q hallucinations: casey@platformer.news and zoe@platformer.news.

Ten strategies for replacing Twitter from people who used to work there

Zoë Schiffer — Fri, 01 Dec 2023 01:00:39 GMT

Behold the glorious cover of Extremely Hardcore

Today, we’re thrilled to share an announcement that has been in the works all year. Zoë’s book about Elon Musk’s takeover of Twitter, Extremely Hardcore, comes out February 27, 2024. And you can pre-order it today.

Extremely Hardcore is the culmination of more than a year of reporting from Platformer on what was perhaps the wildest corporate takeover in the history of Silicon Valley. We think you’ll really like it — but don’t take it from us.

Legendary Bloomberg columnist Matt Levine called it “the definitive book on perhaps the weirdest business story of our time. A fast-paced and riveting account of a hilarious and tragic mess." And Vanity Fair’s Nick Bilton said “Zoë Schiffer’s incredibly written and astonishingly reported book about the Musk era of the world’s craziest company tells the story of a man who took the clown car, strapped a rocket to the back of it, and then slammed it into a wall at 100,000 miles an hour. You simply won’t be able to put this book down.”

Extremely Hardcore is scoopy, juicy, and full of fresh details from the workers left to pick up the pieces after Musk let that sink in. If you’re so inclined, we invite you to snag a hardback copy, which we’re told is helpful in claiming a place on the bestseller lists. We’ll have more to say about the book in weeks to come — but in the meantime, that pre-order link again can be found right here. If you want a signed copy — an exclusive for Platformer subscribers — fill out this form once you’ve pre-ordered.

To celebrate the book’s announcement, today we reached out to a wide range of former employees who used to work at (and, in one case, study) Twitter for their thoughts on how to build its replacement. Threads, Mastodon, Bluesky and others have all gained momentum to varying degrees, but all have a long way to go to reach their full potential.

Where to go from here? Here’s what they told us, lightly edited for clarity and length.

Focus on shipping — Seth Wilson, former director of threat management

You have to constantly innovate and experiment. Speed of innovation is critical in the early stages. In the early years of Twitter, one of the things I loved seeing was new features being rolled out every day. It was fast. So, talk to your engineers and ask if they have the tools and space to experiment and innovate.

You need to establish a culture where good ideas can come from everywhere — from any engineer, any employee, in any line of business. Ideas get filtered out and lost if they pass through a rigid hierarchy. What I loved about Twitter was seeing a junior engineer or intern go up at Tea Time (Twitter’s old weekly all-hands meeting) and demonstrate a feature they’ve developed.

Another source of innovation and ideas came from Hack Week projects. Do not under-invest in this area. There are countless features that Twitter implemented over the years that originated with a Hack Week project. The corporate development team used to run a “TweetTank” competition and have Tweeps pitch a partnership or acquisition that Twitter should do, “Shark Tank” style. One of the winners of that competition wound up being one of Twitter’s most successful acquisitions, Gnip.

Build a great API — Menotti Minutillo, former senior engineering manager for privacy and engineering

History has shown that a well-supported API provides a lot of opportunity and value for content creators, developers, and the platform itself. It's critical for any amount of scale, and helps use cases emerge from experimentation. It made customer service possible. And it allowed for all sorts of wacky automated accounts that gave Twitter a lot of flavor.

I worked on Twitter’s API for two years, specifically on security and privacy features. Once you open up the API there’s the potential for abuse, but it’s not all or nothing. From the beginning, try to establish a system based on developer reputation. That way, developers can earn their way to higher limits or higher degrees of capability. This means disallowing most things by default, and then slowly starting to allow functionality to developers who prove they’re good players — rather than opening it up to everyone and trying to knock down those who display bad behavior.

Also, the more you can have your API mimic the application features, the better. Twitter played catch-up, launching a feature in the mobile app and then having developers be like “when is this coming to the API?” Like when polls came out, it took forever to get it built into the API. Every time you go to general availability with a feature, you also need to have it available in the API. Conceptually it's like “ya of course,” but it’s not an easy thing to do.

Let creators own the audiences they build — Lara Cohen, former global head of marketing and partnerships.

I think creators of all size and scale have realized just how much value they’re bringing to these platforms, how much their content drives engagement and profit — and how little they’re getting in return. To keep creators, platforms need to build with this in mind — think of long-term ways for creators to monetize (i.e., not just launching a splashy creator fund for the headlines and then letting the money run out). Create features that allow creators to own the audience they’ve built there. And build safety features, because often the folks driving the most engagement are also subject to the most harassment.

Draw a clear line on content moderation. — Yoel Roth, former head of trust and safety

You can draw a line that prioritizes safety over speech. Or you can draw a line the way Twitter historically drew it and say, “look, we’re going to prioritize having context, and we’ll have reports, and if you see something say something and sorry tough luck.” You can draw the line anywhere you want! But if you look at expectations of social platforms from 2018 onward, it’s clear that most people believe platforms should moderate proactively, not just reactively.

A key bit of being a public, real-time speech platform is that you’re not moderating one-to-one interactions for the most part. Nor are you even moderating interactions within small, closed groups of folks who’ve chosen to interact with each other. It's a free-for-all. You have to recognize that you lack context on these interactions, and that different people can have different expectations of the same conversation. One person could interpret an interaction as being horrible and offensive. They might be right, or they might be wrong. And you as the platform have only a tiny slice of visibility.

Every platform to date has gotten completely stuck on this. At Twitter, for many years, the operating mindset was “we as a company lack context on these interactions, and consequently need to be pretty hands off.” The result of that was a reporting practice that required a first-person report for an abusive post, which was slanted toward getting the company more context, Like, if the person who was the topic of the post says it was abusive, it’s not just friends fucking around. That’s really hard to do because it puts a lot of burden on the victim of abuse, and it’s out of touch with expectations of social platforms now.

Amplify authentic, positive conversations — Karl Robillard, former global head of social impact.

These bring people together and reduce polarization. Twitter used to invest heavily in this area — bringing company and community together in the spirit of healthy conversation. It paid off in so many ways — building climate emergency response tools, finding missing kids, preserving aboriginal languages, and teaching internet safety and media literacy. When you lead with positivity and goodwill, the world opens up in beautiful and unexpected ways.

Leverage AI to reduce the mental health burden on your content moderators. — Noam Segal, former head of health research

In August, OpenAI published a blog post about using GPT-4 for content moderation. It’s well suited to this sort of work. Policies change frequently, and it can take time for human moderators to learn, understand, and fully implement those changes. Large language models can read the policy and adapt more quickly.

What we’re seeing with AI is that it scales, and humans don’t. And I’m so saddened to say this, but the scale of hatred in this world is unimaginable, and it has to be met with tools that can match the scale of hate.

Since the war in Israel started, I have seen things that I cannot unsee. I, and every other Israeli and Palestinian, am going to need deep therapy to get over this. It’s traumatizing and horrific. As a person who is a parent to three kids, I can’t even speak about the things I have seen.

So I'm afraid of what we are doing to our human moderators. We are exposing people to things that no one should ever be exposed to. Can we afford from an ethical standpoint to put people through this trauma in order to create a healthier discourse? I feel very torn, but I don’t think so.

Don’t be afraid to be opinionated in your content labels — Lisa Young, former head of content design.

Before the 2020 election, our labels on COVID misinformation read “Get the facts from health officials about the science behind COVID-19 vaccines.” The language wasn’t structured, it was hard to localize, and, to conservative users, it reinforced the perception that Twitter was biased. We also tested the word “disputed,” and everyone responded negatively. Finally we landed on the label “misleading” and added a bold caution logo. Then: “Learn why health officials say vaccines are safe for most people.” These labels were easier for our teams to implement, and showed a 17% increase in click-through rate, meaning that millions and millions more people were trusting us to give them more context on potentially misleading tweets.

This project reinforced how important it is to test language, and showed how structured content is essential for creating a scalable, quick-response product.

Roll out a feature like Community Notes to help users govern themselves — Manu Cornet, former software engineer.

When you remove a post outright, you make it easier for the person who posted it to play the victim. Community Notes — which attempt to surface neutral, non-partisan, fact-based clarifications to popular posts that are written by users — offer a powerful alternative to that approach. Elon Musk’s own tweets are regularly flagged by the community. It’s self-regulating in a way.

That said, I don’t know that community moderation is enough. As much as my engineer’s mind wishes most problems could be solved with technology, I’m not naive enough to think algorithms can do 100 percent of this job in a social network. The community can help debunk fake stories, but many other content moderation challenges need humans, at least right now.

Add lists! — Shauna Wright, former senior content strategist.

I want to be able to track NBA reporters, political reporters, and cultivate my specific interests like I did on Twitter, without having to follow those accounts and always see them in my main timeline. Like, I’m a Warriors fan, and when there’s a game on I want to see what people are talking about. But I don’t want all those tweets in my timeline all the time.

Take brand safety seriously — Julianna Hayes, former senior vice president of sales and corporate finance.

Monetizing social networks is a challenge. From a revenue product perspective, find something that feels as organic to your platform as possible. Your ad products will need to drive results. Advertisers need to find value in their performance; no one owes you their spend. You will have to work hard for each sale, as you are up against companies that have products that perform exceptionally well. A focus on brand safety is key, as advertisers' reputation is everything. Have that in the back of your mind as you design advertiser tools and safety measures.

And here’s a bonus tip from someone who didn’t work at Twitter but that every would-be Twitter replacement needs to have on their mind, from one of our favorite thinkers in the space.

Get ready for a flood of new regulations around the world. — Evelyn Douek, assistant professor of law at Stanford Law School, co-host of the Moderated Content podcast, and someone who was really good at tweeting back in the day.

For a long time these platforms were unregulated. But we’re seeing in the last couple years a raft of new regulatory requirements coming in, like the Digital Services Act in the EU, and the Online Safety Act in the UK. A bunch of different states are passing different bills in the United States, and some are on appeal in the Supreme Court. The writing on the wall is clear: governments everywhere are getting way more active in regulating these technologies, and new platforms are going to have to deal with that much more than they did in the past.

On the podcast this week: Kevin and I sort through a week’s worth of news about OpenAI. Then, the Times’ David Yaffe-Bellany joins to remind us why CZ is going to prison. And finally, the latest news about how AI sludge is taking over web search.

Governing

A US judge blocked Montana from implementing its attempted ban of TikTok on Jan. 1, citing free speech concerns. (David Shepardson / Reuters)
The CEOs of TikTok, X, Meta, Snap and Discord are set to testify at a Senate hearing on online child sexual exploitation in January. (David Shepardson / Reuters)
Meta is challenging the constitutionality of the FTC’s in-house courts amid potential restrictions on how the company can monetize user data. (Jan Wolfe / The Wall Street Journal)
The Canadian federal government reached a deal with Google over the Online News Act, where Google can continue to share news online by paying news companies annually. Another successful shakedown. Which countries will be next to line up at the trough? (Daniel Thibeault, David Cochrane and Darren Major / CBC)
Apple will have to face a revived probe into its dominance in mobile browsers and cloud gaming by the UK Competition and Markets Authority, after the agency won an appeal. (Katharine Gemmell / Bloomberg)
The rules for political and social issue advertising for the 2024 elections will be the same as past election cycles on Facebook and Instagram, Meta says. That means no political ads starting a week before the election. (Anna Edgerton and Alexandra Barinka / Bloomberg)
After finding out their images were being used for AI nude deepfakes, a group of women in New York are fighting back and calling for more regulation. (Olivia Carville and Margi Murphy / Bloomberg)
The Data and Trust Alliance, a consortium of large companies, has developed a standard for describing the origin, history, and legal rights of data. (Steve Lohr / The New York Times)
A dozen big tech companies signed onto the UK Online Fraud Charter, which aims to fight online scams, fake ads and romance fraud. (Ben Mitchell / The Independent)
Adobe is reportedly rushing a proposal to address European regulator concerns over its Figma acquisition. The proposal could include not tying Figma to Creative Cloud and divesting Adobe XD, which competes with Figma. (Samuel Stolton, Katharine Gemmell, Leah Nylen and Brody Ford / Bloomberg)
As TikTok begins work on its Norwegian data center, the company is pledging €12 billion for the next 10 years to appease European regulators. (Paul Sawers / TechCrunch)
Palestinian creators are finding it difficult to access YouTube’s revenue sharing and other Google services amid the war in Gaza. (Paresh Dave / WIRED)
A new wave of appeals to Meta’s Oversight Board related to content moderation around the Israel-Hamas conflict could reshape moderation policies. Or the board could take three years to hear the cases and Meta could ignore the eventual policy recommendations. (Russell Brandom / Rest of World)
Foreign governments, particularly in Russia, Iran and China, are likely to continue pushing influence campaigns through fake social media accounts in 2024, Meta warns. Meanwhile, Meta reported that the US government stopped sharing data with it in July — likely amid legal uncertainty created by the “jawboning” cases now before the US Supreme Court. (AJ Vicens / CyberScoop)
China has “massively increased” the amount of cyberattacks on Taiwan in the last six months, Google cybersecurity experts say. (Ryan Gallagher / Bloomberg)

Industry

Threads is reportedly launching in Europe in December, its largest market expansion since launch. (Salvador Rodriguez, Sam Schechner and Meghan Bobrowsky / The Wall Street Journal)
- Keyword search on Threads is now widely available and will soon support “all languages”, Instagram head Adam Mosseri says. (Sarah Perez / TechCrunch)
Meta paused shipments of the Quest 3 Elite Battery Strap after user reports of a charging fault that made the battery useless. (Scott Hayden / Road to VR)
Startup Stability AI is reportedly exploring a sale as it faces investor pressure over its financial position. People give OpenAI a lot of grief these days but Stability is arguably even messier. (Mark Bergen and Rachel Metz / Bloomberg)
Elon Musk wants X’s fleeing advertisers to “go f— themselves”, claiming that the companies were blackmailing him with advertising. He singled out Disney’s Bob Iger. (Lora Kolodny / CNBC)
- But he says his antisemitic post was a “mistake” and possibly “the most foolish” thing he’s done on X. (Jacob Kastrenakes / The Verge)
- Meanwhile, advertisers are reacting to Musk’s latest meltdown exactly how you’d expect: by pledging to never advertise again. (Kate Conger / New York Times)
OpenAI reportedly doesn’t plan on including outside investors, including Microsoft, on its new board of directors. (Amir Efrati, Jessica E. Lessin and Aaron Holmes / The Information)
- But Microsoft is getting a non-voting observer seat on the board. (Alex Heath / The Verge)
- Some high-profile women in AI say they would not consider joining the current all-male board either for fear of being marginalized. (Kate Knibss, Lauren Goode and Khari Johnson / WIRED)
- Despite playing a part in Sam Altman’s firing and his CEO position at Quora, which has become increasingly competitive with ChatGPT, Adam D’Angelo remains on the OpenAI board. (Priya Anand and Sarah McBride / Bloomberg)
- In an interview, Sam Altman said he was initially hurt and angry when he was ousted, but declined to say why he was fired. (Alex Heath / The Verge)
Adam Selipsky, Amazon’s cloud division head, says that companies “don’t want a cloud provider that’s beholden primarily to one model provider,” in a shot at OpenAI. (Camilla Hodgson and Tim Bradshaw / Financial Times)
Researchers found that ChatGPT’s training data can be leaked through a “divergence attack” — asking the chatbot to repeat a word constantly. (Alex Ivanovs / StackDiary)
Amazon introduced its AI chatbot for companies, Q, built to be more secure and private than other chatbots. (Karen Weise / The New York Times)
- The company also released the AI-powered Titan Image Generator, which can create new images or modify existing ones. (Kyle Wiggers / TechCrunch)
TikTok launched artist accounts to improve engagement and discoverability as the company pushes further into streaming. (Sheena Vasani / The Verge)
Apple, the company that produced Robert De Niro’s new film, altered his speech last minute at the Gotham Awards to take out criticisms of Donald Trump and focus more on the film. (Brent Lang and Matt Donnelly / Variety)
Advertisers running Google search ads are automatically opted in to the Google Search Partners Network, where their ads often run on controversial third-party websites, a report found. (Natasha Lomas / TechCrunch)
Google Registry released a new top-level domain, “.meme. (Emma Roth / The Verge)
Google fixed its sixth zero-day exploit this year through an emergency security update on Chrome. (Sergiu Gatlan / Bleeping Computer)
Researchers at Google DeepMind used an AI tool to discover 2 million crystal structures, opening up possibilities in renewable energy and advanced computation. (Michael Peel / Financial Times)
Gmail accounts that haven’t been used in two years will soon be deleted under a new policy. (Dalvin Brown / The Wall Street Journal)
Google is celebrating 1 billion monthly active users for RCS messaging by and adding “Photomojis,” which let users make emojis out of photos, along with a slew of other new features. (Allison Johnson / The Verge)
The latest Android updates include new emoji features, emoji for voice messages, and an AI-generated image description tool. (Lawrence Bonk / Engadget)
Users can now hide locked WhatsApp chats behind a customizable secret code. (Emma Roth / The Verge)
This year’s Spotify Wrapped is out, with Taylor Swift as the most-streamed artist and new features such as listening preference city matching and streaming habit highlights. (Ann-Marie Alcántara / The Wall Street Journal)
- Some users who were city matched to Burlington, Vt., Cambridge, Mass. or Berkeley, Calif. are joking that those cities were designated for LGBTQ users. This story was highly relevant to me, an LGBT Spotify user whose Wrapped located him in Burlington (Madison Malone Kircher and Sopan Deb / The New York Times)
San Francisco startup Perplexity introduced two new online large-language models with real-time data that can provide updated responses. (Kristi Hines / Search Engine Journal)
Mailchimp is ending its newsletter service TinyLetter to focus on its core marketing product. Which sucks. (Jay Peters / The Verge)
Substack is rolling out a suite of video content creation tools, putting it more in direct competition with Patreon. (Taylor Lorenz / Washington Post)
Pinterest is testing a “body type ranges” tool to search on the platform, in an effort to boost inclusivity. (Sarah Perez / TechCrunch)

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

The AGs reveal their hand against Meta

Casey Newton — Wed, 29 Nov 2023 01:02:51 GMT

Meta’s headquarters in Melo Park. (Tayfun Coskun / Getty Images)

On October 24, 41 states and the District of Columbia sued Meta. Attorneys general alleged that the company hurts younger users by intentionally designing addictive and harmful products at the expense of their privacy and well being.

At the time, the lawsuits’ merits were difficult to assess, since much of the evidence in the 233-page complaint was redacted. But this week, with Meta’s approval, the full complaint was made public. And what we could only speculate about last month now seems ripe for a fuller discussion.

Writing about the subject last month, I divided the complaint into two big parts. One is focused on privacy; the other on everything else. Let’s take privacy first.

The Children’s Online Privacy Protection Act (COPPA) places several requirements on platforms that collect data on users under the age of 13. Chief among these is the need to obtain verifiable parental consent for the user; it also puts restrictions on how data on younger users can be collected and used.

In practice, all of this is more trouble than it’s worth. And so most major platforms prohibit under-13s from signing up, while remaining well aware that underage users are using their services daily. This means platforms get to have their cake and eat it too: reap the rewards of the engagement and ad revenue that comes with younger users on the platform, while also insisting that they’re not there, or that if they are there they shouldn’t be, and also look at how many under-13 accounts we remove every year.

Platforms’ wink-wink, nudge-nudge relationship with COPPA has been mostly tolerated until now. But the unredacted lawsuit lays out what appears to be a straightforward case of a company well aware it is collecting reams of data on underage users and not seeking parental permission as required.

The lawsuit alleges:

Within the company, Meta’s actual knowledge that millions of Instagram users are under the age of 13 is an open secret that is routinely documented, rigorously analyzed and confirmed, and zealously protected from disclosure to the public.
Meta’s extensive internal records documenting its actual knowledge of its under-13 Instagram users and collection of data from those users include the following: (1) charts boasting Instagram’s penetration into 11- and 12-year-old demographic cohorts; (2) an internal report presented to Zuckerberg regarding the four million under-13 users on Instagram; (3) emails and policies documenting Meta’s mishandling of known under-13 user accounts; (4) discussions among Meta’s researchers taking pains to avoid uncovering Instagram’s under-13 users through their studies; (5) documents admitting that Instagram’s registration process regularly elicits false self-reported ages from its under-13 users; and (6) data from Meta’s age-estimation algorithms confirming that millions of individual Instagram accounts belong to children under the age of 13.

Let’s stipulate that age verification is difficult, controversial, and has even been held in some cases to be unconstitutional. Forcing children to verify their ages can also mean placing undue burdens on adults. (Which frankly seems to be a primary objective in the spate of anti-porn bills passed by legislatures this year.)

I’ve talked with several trust and safety executives this year who are all but begging for a workable solution to age verification, which will likely involve linking your online life to some trusted authority who knows your true age and silently passes that information along as a token to the apps on your devices.

It’s no coincidence that two weeks ago, in the run-up to the release of the unredacted lawsuit, Meta called on the two biggest app store operators — Apple and Google — to assume responsibility for age verification at the device level. On the most recent episode of Moderated Content, Alex Stamos called for a modified version of this approach that would allow parents to input their children’s ages upon setting up their devices. All of these approaches are worth discussion, and arguably all are better than the current system.

In the meantime, though, the law is the law. And it looks like the attorneys general have Meta dead to rights.

I ran all this by Meta, and the company shared a statement beginning “Instagram’s Terms of Use prohibit users under the age of 13 (or higher in certain countries) and we have measures in place to remove these accounts when we identify them.” It did not address any of the specific claims or evidence in the unredacted lawsuit, but it did once again say that there should be a law making all of this the responsibility of Apple and Google.

II.

That leads to the second, more dramatic set of allegations against Meta: that it engaged in a “scheme to exploit young users for profit,” routinely ignoring its own research on product changes that would improve the well being of younger users for fear of losing out on revenue and engagement.

When the redacted lawsuit appeared, the public claims related largely to Meta’s use of push notifications and other strategies to get young people to open their apps repeatedly. This can be annoying, I wrote at the time, and on the margins can encourage overuse of the apps. But it has also become the industry standard for all kinds of apps, and it was not clear to me how the AGs could get from there to a conviction.

The redacted lawsuit adds significant evidence to the claims. By focusing on a handful of key battles inside the company, it identifies cases where Meta found useful strategies for improving user well being and then chose not to implement them for business reasons.

The most fascinating of these case studies might be the lawsuit’s look at Project Daisy, Instagram’s much-ballyhooed experiment with hiding like counts for public posts. After two years of study and user testing, in May 2021 the company announced it would allow users to hide their likes — but would make the feature opt-in rather than opt-out. Hiding likes had no clear effect on well-being, we were told at the time, and wouldn’t it be better to give users the choice?

The AGs lawsuit alleges that Meta lied to us about the conclusions it had drawn. Visible like counts can heighten what the company calls “negative social comparison” — the feeling of inferiority that comes from seeing friends, family members, and influencers whose lives appear more glamorous, and whose posts are receiving more engagement than your own. That visible like counts would make many people feel worse about themselves is intuitive, and yet when Meta discussed the issue in 2021 it effectively threw its hands up and said there was no making sense of how likes made people feel.

In reality, the AGs found, the company had documented the effect that visible likes has on well being with admirable precision. They write:

A January 2020 study that Meta conducted noted that “[s]ocial comparison is . . . higher among younger than older people. Younger people are more susceptible to peer influence and social comparison.”
Just three months later, Meta conducted another internal study that revealed that:
“About 1 out of 10 people experience negative social comparison on Instagram often or always” (emphasis in original);
“About 1 in 4 people think that Instagram makes social comparison worse”;
“People receive about 5% as many Likes on their own posts as those they see on IG”; and
“Seeing high Like counts is associated with feeling worse (more negative, less positive comparison).”
Researchers at Meta documented that appearance-based comparison on Instagram was worse for teenagers and young adults compared to older adults, and worse for female users across all age groups 60 and under. The highest rates of negative appearance-based comparison on Instagram were for teen girls, aged 13 to 18.
In an internal presentation from 2021, Meta employee noted that teens suffered from “constant negative comparisons” on Instagram because Meta recommended triggering content to them and continued showing them Like counts.

Moreover, the company found, Meta is aware that “66% of teen girls on IG experience negative social comparison,” along with “40% of teen boys.”

This week, I talked to someone who used to work on these issues at Meta who told me that when the company rolled out the neutered, opt-in version of Project Daisy, it knew that doing so would have negligible effects on well being.

But the non-neutered version did decrease revenue around 1 percent, the AGs report, because despite the harm it causes, people really do like browsing Instagram to see how many likes posts get.

Asked about all this, the company repeated what it told me in 2021: that results from Project Daisy were inconclusive.

A similarly compelling case study comes with the AGs recounting of an internal battle over augmented reality beauty filters. In 2019, the company debated whether to permit AR filters that simulate plastic surgery. After an outcry over the potential effect of these filters on young girls in particular, the company temporarily banned them.

After consulting with global experts on well being, the company’s head of responsible design argued that Meta should ban the filters permanently. “When it comes to products or technology that are used extensively by minors (under 18), I do believe we have an obligation to act more proactively in mitigating potential harm,” she wrote. Instagram’s head of policy agreed that the ban should be extended, noting that the filters are “actively encouraging young girls into body dysmorphia and enabling self-view of an idealized face (and very western definition of that face by the way) that can result in serious issues.”

In the end, though, CEO Mark Zuckerberg overturned the ban, saying he found it overly “paternalistic,” according to the lawsuit. The company did later approve more limited restrictions on the filter.

“While filters exist across every major social platform and smartphone camera, Meta bans those that directly promote cosmetic surgery, changes in skin color or extreme weight loss,” the company told me.

There’s much more in the lawsuit; for further details I recommend reading Jeff Horwitz’s piece in the Wall Street Journal and Devin Coldewey’s in TechCrunch.

As distasteful as some of these anecdotes are, it’s still uncertain whether the AGs will be able to prove their claim that Meta knowingly built addictive and harmful products. The “knowingly” aspect is famously hard to prove, and after the release of Frances Haugen’s Facebook Files, the company wound up dismantling many of the research projects that would have led it to know about potential harms in advance.

But is there evidence here that in at least some cases, the company pursued growth and revenue at the clear expense of user safety? It sure seems that way to me. A month ago, it was possible to believe that the AGs may have overplayed their hand. Today, I’d say Meta is in for a very tough fight.

This article has been updated to reflect that it is the original version of Project Daisy that was found to reduce revenue.

Governing

The UK’s Competition and Markets Authority provisionally found Adobe’s acquisition of Figma will likely harm innovation in software. I love the Figma people, but is there any argument that the acquisition isn’t going to make the cost of Creative Cloud go up? Who are the other challengers on the horizon? (Joseph Hoppe / The Wall Street Journal)
There’s no widespread harm to mental health from internet use, including social media and gaming, a group of researchers at the Oxford Internet Institute found, though more data was needed from tech companies to confirm a connection. (Tim Bradshaw / Financial Times)
Teenagers and children in the UK are more likely to embrace generative AI than adults, new research shows, with Snapchat’s MyAI being the most popular tool among younger people. (Ofcom)
While Telegram has “restricted” some extremist channels, including two used by Hamas, content shared within those channels is still accessible to the public. (Vittoria Elliot / WIRED)
An increasing number of white-supremacist, neo-Confederate, and explicitly Nazi newsletters are being hosted on Substack, taking advantage of its lax content moderation policies. The correct number of newsletters using Nazi symbols that you host and profit from on your platform is zero. Here’s hoping Substack takes decisive action here, which would be consistent with their written anti-hate speech policies. (Jonathan M. Katz / The Atlantic)

Industry

Some major advertisers have not only paused their campaigns on X, but also stopped posting on the site altogether, reportedly due to concerns around brand safety. (Oliver Darcy / CNN)
- Elon Musk posted a meme on X supporting the far-right conspiracy theory Pizzagate, that alleged Democratic leaders ran a satanic child sex ring out of a pizzeria in D.C. He later deleted it. (Drew Harwell / Washington Post)
Google DeepMind’s chief scientist, Jeff Dean, discusses strategies to combat misinformation fueled by AI, preventing hallucinations, and Google’s cautious approach to generative AI. (Samidha Sharma / The Economic Times)
Google’s first geothermal project, which aims to power data centers with clean energy, is up and running in Nevada. (Justine Calma / The Verge)
Sports Illustrated reportedly published articles that were attributed to AI-generated authors, complete with AI-generated headshots. The bylines have since changed its attribution to “a 3rd party.” This kind of sludge is about to make up 80 percent of the internet, just watch. (Mia Sato / The Verge)
- See also David Roth at Defector on this debacle.

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

The OpenAI saga isn’t over just yet

Casey Newton — Tue, 28 Nov 2023 01:00:37 GMT

(Jaap Arriens / Getty Images)

On the surface, the OpenAI team that showed up to work today looks almost identical to the one that showed up on November 17, the morning that CEO Sam Altman was suddenly and shockingly fired, roiling the tech world and leaving the company’s future in question. Ten days later, Altman is back, along with company president Greg Brockman and the hundreds of employees who threatened to resign if OpenAI’s board would not reverse its decision. Viewed from a sufficient distance, it appears odd that the net result of Silicon Valley’s most engrossing drama in recent memory was a barely modified status quo.

The conventional wisdom now holds that OpenAI’s nonprofit board overplayed its hand, communicated its decision and motives terribly, and disqualified itself from governing the most important company of its generation. And I basically agree with all that, as I wrote here last week. Whatever reasons the board may have had for declining to outline with any specificity why they fired the CEO who had led their organization to great success, in the end their silence doomed them.

And yet: for everything they failed at, the board does appear to have succeeded in putting a new governance structure into place. Altman and Brockman are no longer on the board; Quora CEO Adam D’Angelo, who voted to fire Altman, remains as well, and will help to fill out the rest of its membership. Bret Taylor, the former Twitter board chair, and Larry Summers, the former US Treasury secretary, will join D’Angelo on the board; together, they will appoint up to six new members. In addition, the new board will commission an independent investigation into the events surrounding Altman’s firing.

Corporate investigations vary widely in quality and rigor, and it remains to be seen whom the board will ask to conduct this one. There’s no guarantee that the results of the investigation will be made public, though there’s a case to be made that they should be, at least in part. And whatever the investigation finds, it’s not clear at this point what it would have to find to justify Altman’s removal — especially to the company’s employees, whose near-unanimous support for their CEO is all but unheard of in Silicon Valley.

Still: it feels too early to declare the story over. The soon-to-be-former members of OpenAI’s board will speak eventually: to investigators, the public, or both. Squads of investigative reporters are now digging into Altman’s sprawling web of investments.

It is quite possible that a few years from now, the events of this month will be a footnote in the history of OpenAI’s ascent. But with a new board arriving and an investigation about to begin, I wouldn’t be surprised if this story didn’t have another twist or two in store.

I spent the past several days chatting with sources close to OpenAI and its board. Here are a few items I can report that may bring some additional texture to the past week’s events.

Once the board determined that it had a majority of members willing to fire Altman, it felt a pressure to act quickly. The board appears to have anticipated correctly that Altman would muster enormous resources to prevent his removal from the company. Once it determined it had a majority of members willing to fire Altman, it sought to move before he got wind of the decision.

But in the board’s haste to fire him, it failed to plan for everything that would follow: starting with employees’ utter incredulity at what was happening, and the minimal explanations that would be offered to support the board’s decision. Moving quickly may have meant that the board was able to fire Altman before he could stop it, but it also created the conditions for his return.

Still, Altman got a worse deal on Tuesday than he did the day he was fired. I’m told that Altman’s team originally called for him to be reinstated and every current member of the board to resign. By holding out for a few days, the board did get some concessions: D’Angelo remaining on the board; the outgoing board members getting input on members of the new board; and the independent investigation. It’s not nothing.

Bret Taylor had been considered for OpenAI’s board before. OpenAI had been seeking new board members ever since three people stepped down from it earlier this year; one mystery has been who was under consideration. I can report that one person the board had talked to was Taylor, a highly regarded entrepreneur and board operator who also served a stint as co-CEO of Salesforce. But the board had been unable to come to a consensus on any new members before the firing, I’m told.

The board never received any formal communication about Q*. One of the more intriguing stories about the drama to come out over the past few days concerns Q* (pronounced “Q-star”), an AI model that can solve basic math problems. The Information (which has really done outstanding work on the whole OpenAI story) reported that Q* “raised concerns among some staff that the company didn’t have proper safeguards in place to commercialize such advanced AI models.”

That story followed a report from Reuters that said “several staff researchers wrote a letter to the board of directors warning of a powerful artificial intelligence discovery that they said could threaten humanity.” I can report that the board never received any such letter about Q*.

The board never received the letter that Elon Musk posted, either. Last week a letter purporting to be from OpenAI staffers briefly appeared on GitHub. Like the board’s message in firing Altman, it was notably short on specifics. “Throughout our time at OpenAI, we witnessed a disturbing pattern of deceit and manipulation by Sam Altman and Greg Brockman, driven by their insatiable pursuit of achieving artificial general intelligence,” the letter read.

Musk posted a link to the letter on X, along with the comment “These seem like concerns worth investigating.”

In any case, I’m told, no such letter was ever received by the OpenAI board.

How Will AI Affect the 2024 Election?

Tuesday, November 28, 6–7 p.m. ET

2024 will bring the first presidential election of the generative AI era. As artificial intelligence produces output that is increasingly difficult to distinguish from human-created content, how will voters separate fact from fiction? The Brennan Center for Justice and Georgetown University’s Center for Security and Emerging Technology are convening experts to examine these and other critical questions about how AI might impact election security, voter suppression, election administration, and political advertising and fundraising.

Join the Brennan Center and CSET for this live virtual panel, which will explore what steps the government, the private sector, and nonprofits should take to minimize the possible dangers while harnessing the benefits of these new and powerful tools.

RSVP TODAY

Governing

A district judge ruled that the FTC can continue to revise Meta’s 2020 privacy settlement, blocking a move from Meta to force the issue into court. (Leah Nylen / Bloomberg)
A newly unredacted lawsuit by 33 state attorneys general alleges that Meta knew of millions of Instagram users under the age of 13, but only disabled a fraction of those accounts and continued to collect data on children. (Natasha Singer / The New York Times)
- A test of Instagram Reel’s algorithm found that accounts following children were served a mix of videos including risqué footage of children and overtly sexual adult videos. (Jeff Horwitz and Katherine Blunt / The Wall Street Journal)
A group of 18 countries, including the US and UK, have set up an international agreement on how to keep AI safe from bad actors, urging companies to make systems “safe by design”. (Raphael Satter and Diane Bartz / Reuters)
Amazon was issued a fine of just $7,000 — the maximum penalty — after an employee died in a distribution center in Indiana, underscoring the need for stronger penalties and safety regulations. (Caroline O’Donovan / Washington Post)
Amazon is reportedly set to get unconditional approval from the European Commission for its $1.4 billion acquisition of robot vacuum maker iRobot. (Foo Yun Chee / Reuters)
The UK chancellor’s Autumn Statement announced that the government will boost spending for computing power to develop AI and UK-based quantum computers aimed at running operations without errors. (Clive Cookson / Financial Times)
An analysis found that ads from 86 major advertisers on X were on viral posts spreading misinformation about the Israel-Hamas conflict, with both X and creators sharing in ad revenue. (Jack Brewster, Coalter Palmer and Nikita Vashisth / NewsGuard)
- X could lose up to $75 million in advertising revenue as more advertisers pause their campaigns on the platform after Elon Musk endorsed an antisemitic post. (Ryan Mac and Kate Conger / The New York Times)
Israel told Musk that his Starlink satellite network will only operate in Gaza if it gets approval from the Israeli Ministry of Communications. Musk visited the country in a goodwill gesture after he was blasted for supporting antisemitic comments on X. (Chloe Cornish / Financial Times)
Some parents are being mislabeled as child abusers by Google’s AI-powered systems that review YouTube content, leading to unnecessary investigations and financial burdens. (Kashmir Hill / The New York Times)
A slew of lawsuits against AI companies over copyright infringement, all led by an unlikely new attorney named Matthew Butterick, could shape the future of AI and creative industries. (Kate Knibbs / WIRED)
The emergence of uncensored AI models and chatbots is sparking questions about balancing safety and freedom in AI. (Mark Gimein / The Atlantic)
Russia‘s interior ministry has reportedly added Meta spokesperson Andy Stone to its wanted list. A reason was not indicated. (Laura Kayali / POLITICO)
After Chinese social media companies like Weibo introduced a new rule that required influencers to display their legal names on their profiles, influencers are removing followers or quitting altogether. (Caiwei Chen / Rest of World)

Industry

Reddit is reportedly in talks for a potential initial public offering as soon as the first quarter of 2024. (Amy Or, Ryan Gould, Katie Roof and Gillian Tan / Bloomberg)
Content moderators for dating apps say there are high workloads, unrealistic targets, and a lack of mental health support, putting their well-being and user safety at risk. (Niamh McIntyre / The Bureau of Investigative Journalism)
ByteDance is reportedly planning on cutting hundreds of jobs in its gaming department and winding down Nuverse, withdrawing from the gaming industry. (Zheping Huang and Dong Cao / Bloomberg)
TikTok’s research on the platform’s impact on the music industry found that users are more willing to pay for a streaming service or music product than the average consumer. (Stuart Dredge / Music Ally)
X preview cards will get headlines again, Musk says, after removing them last month in a bit to discourage people from leaving the site. (Ivan Mehta / TechCrunch)
Google Bard is now better at understanding YouTube videos, including the ability to ask Bard specific questions related to video content. (C. Scott Brown / Android Authority)
Google Meet users can now “raise hand” in a meeting by physically raising their hands instead of clicking a button. (Abner Li / 9to5Google)
Google said that a missing file problem in Drive was caused by the desktop app, and is investigating the problem. (Ben Schoon / 9to5Google)
Instagram is now letting all users download publicly posted Reels. (Andrew Hutchinson / Social Media Today)
A look at the effective altruism movement, which set out to encourage better outcomes in philanthropy but has struggled in high-profile roles at FTX and OpenAI. (Robert McMillan and Deepa Seetharaman / The Wall Street Journal)
Startup Inflection AI unveiled a new AI model that it says outperforms Google’s PaLM Large 2 and Meta’s LLaMA 2, and will soon be integrated into its chatbot Pi. (Alex Konrad / Forbes)
China’s Douyin has gained popularity among older generations, and has become an outlet of self-expression and connectivity for the elderly. (Lavender Au / WIRED)

Those good posts

(Link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Why one AI executive quit his job to protest creators' rights

Zoë Schiffer — Wed, 22 Nov 2023 01:00:28 GMT

Ed Newton-Rex

Programming note: Platformer will be off Thursday for the Thanksgiving holiday.

One of the fiercest debates in artificial intelligence — perhaps second only to who should be CEO of OpenAI — is whether and how creators should be compensated when their work is used to train models for generative AI. For the most part, today’s most popular large language models have been created without the consent of the people whose work now powers them. And arguments about the issue have now spilled over into lawsuits.

Earlier this month, AI companies, including StabilityAI, Anthropic, Meta, Google, and Microsoft, submitted comments to the US Copyright Office on potential rules that would govern how corporations use copyrighted work. While the companies offered varying opinions, the core of their argument was that AI companies should be able to train LLMs using copyrighted works without compensating their creators, according to Wes Davis at The Verge.

Perhaps unsurprisingly, many creators disagree. But to date, they have struggled to gain much ground in court. Today, a federal judge dismissed most of the claims made by comedian Sarah Silverman in one such lawsuit against Meta.

Here’s Winston Cho at the Hollywood Reporter:

U.S. District Judge Vince Chhabria on Monday offered a full-throated denial of one of the authors’ core theories that Meta’s AI system is itself an infringing derivative work made possible only by information extracted from copyrighted material. “This is nonsensical,” he wrote in the order. “There is no way to understand the LLaMA models themselves as a recasting or adaptation of any of the plaintiffs’ books.”

But while they have struggled to convince judges, creators won support in surprising place last week. Ed Newton-Rex, who was the vice president of audio at Stability AI, abruptly resigned from his job. In an op-ed, Newton-Rex wrote that he disagreed with the company’s stance on fair use. “Companies worth billions of dollars are, without permission, training generative AI models on creators’ works, which are then being used to create new content that in many cases can compete with the original works,” he wrote.

On Friday, Newton-Rex spoke with Platformer about why he left, what the ideal arrangement is between AI companies, and how much he expects to make from his own musical efforts.

Zoë Schiffer: You recently resigned from your role as the VP of Audio at Stability AI, and published an op-ed saying that you disagree with the company's position that AI development should be protected by fair use. Can you elaborate?

Ed Newton-Rex: It’s funny in a way. I resigned from Stability, yes, but the focus for me isn't really on Stability so much. It's on all of the large generative AI companies, which have all adopted the same approach. And that approach is that using copyrighted content to train generative AI systems is okay because it's a fair use of that copyrighted content. So it's that position that's held by the general industry that I disagree with. It almost feels like I've resigned from that bit of the industry — not Stability, which does, to be fair, have some nuanced views on this.

In reading your letter I thought what you were saying is that you approved of training LLMs on copyrighted material, but you disagreed that the work that came out of that should be copyrighted.

No, it's actually kind of the opposite. It’s really about what goes into it. So the way lots of these models are trained is obviously just on lots and lots of data. And it's always been really important to me that companies building these models work with creators, and ultimately only work with data that is provided with the consent of the people who own that data and who created it.

To me, there's both a fair-use argument and an abstract moral argument, and both of them say to me that you should really get consent from the people who create the work in order to train on it.

And how were those conversations and debates playing out when you were at Stability?

Well, so I led the audio team, and we actually released a music generation model called Stable Audio a couple of months ago. And that was trained on licensed data where we partnered with a big stock audio library called Audio Sparks. There was a revenue share model and there was an opt-out for musicians, and some of them opted out. That felt to me like a really good way of doing this. Now obviously, I think what Stability has trained other models like images on is publicly documented and is obviously a different approach.

So I don't necessarily want to get into the details of internal conversations like who said what when, but ultimately, I think you just have to look at the company. Every company has an official position on this kind of thing. And in this case, my view does not align with that official position. And again, really, I know I just keep saying this, but I do want to stress that I would've had to resign from a bunch of other generative AI companies at about the same time had I happened to be working there, because I think a lot of them put in these statements that I really disagree with.

Whoever did your media training did a good job. You're very diplomatic.

I don't know about that. I'm not sure. I'm not sure my wife would say I'm the most diplomatic person.

Is it fair to say that those who disagree with you, either in the industry or at the company, feel that the approach that you took with the Stability AI audio product isn't scalable?

There's definitely this narrative in the industry that these approaches don't work and aren't scalable. I think that's probably wrong. There are a bunch of companies and products who are taking these kinds of approaches and actually quite successfully. Stable Audio itself was received really well. It was one of Time's best inventions of the year, which was super cool. And in general we had really, really positive feedback. And I generally think if you work hand in hand with creators, it's a pragmatic thing for an AI company as well, it’s going to make you look good.

There are a number of companies experimenting with different approaches. So there's an interesting one in the image space that's called BRIA, based in Tel Aviv, and they've got an image generation model and they work directly with rights holders. And they actually have a system for whenever you generate an image, to basically figure out which rights holders to pay out accordingly based on whose images had the most bearing on that outputted image. And that's working well for them. They've got a number of clients, I think, and it's going well.

And ultimately, that's the thing I want to see more of. I just want to see more people experimenting with this and saying, yeah, we should be working with and getting consent from creators and let's try some different models to get there. It'll slow us down a bit. Sure. It'll cost a bit more, but ultimately, I think it's the right thing to do.

You’re a musician. I'm curious how that is informing your thinking on this issue.

It informs my thinking quite a lot. Funnily enough, I've worked in music AI generation for a long time, for 13 years. And most people I know in the space, both in my team at Stability and in my old company Jukedeck, are musicians. And I'm not sure that's the case in image generation or video generation or all these other modalities. And that might be why, I think in general, you have more music AI companies who are working with creators as opposed to citing fair use.

On a personal level, that certainly affects my thinking. I'm a composer. I have a bunch of music out there. I don't like the idea that a generative AI system could be set up by a commercial company and could make money by training on my creative output without any kind of say so from me at any point, without me being even asked about that, whether or not I'm being paid.

For me, it comes down to the question of replacement of human work and human creativity. AI companies will focus on the fact that the technology is there to assist creativity. And it's true, it’s a very helpful thing.

Music generation, you can get it to inspire you to give you ideas, you can get it to complete things you're doing, you can get it to create variations of what you're doing. There's loads of assistive technology. The problem is the same models can already be used to generate full, essentially human level creative output. You can easily imagine someone just generating lots of music, uploading it, creating a new stock audio library. And if you do that, then that's clearly affecting the market for the original work. And that is my concern.

And that's one of the factors that comes to bear on the decision of whether something is fair use, and why I don't think this practice is fair use. One of the factors that comes to bear is what is the effect on the market? And it seems pretty clear to me that because these systems are getting so good now, maybe this wasn't the case a few years ago, but now they're so good that I think you can easily argue that there is a negative effect on the market, the original. And so I don't think it's fair use, accordingly.

You said in your TED Talk that machines being creative will never stop humans from being creative. Has your thinking on that changed? Where does the question of copyright come in?

Good find (laughs) — that was a while ago. I look very young. For what it's worth, it's true in that I don't think we'll ever stop being creative. I think there’s a good chance that actually people spend more time being creative in the future, perhaps not for the best reasons. I'm a little worried about artificial general intelligence and what it does to the labor market in general across all industries. And I think that could, maybe unfortunately, end up giving people more free time to be creative. I also think that AI is actually going to really help people be creative in many, many ways. I think it's going to be a huge thing in music education. I think lots more people will be able to learn music because you will essentially have personalized music educators at your fingertips. Whereas at the moment, learning music is kind of a pretty elitist thing.

So I don't think machines will ever replace human creativity in that way. What I am worried about, and the reason that I guess I'm speaking up on this, is the impact on the commercial market for human creativity. We're getting to the stage where all of these models are incredibly good. And I think that inevitably and unavoidably, they’re detrimental to entry level jobs, the things that give people their start, the things that let creators support themselves at the start of their career. These are industries that are going to be dramatically changed by generative AI, and I don't think it's necessarily the right thing to do to try to stop that. There will be lots of advantages to this technology as well. And industries do change.

I think the one thing that I am set on right now is the fact that if you're working towards that, you have to be doing it in the right way. If you're working towards this new technological paradigm, you can't be doing it in a way that is essentially exploitative, is taking advantage of people's creative output unfairly and ultimately putting them at a disadvantage in the process. So that's why I think this particular question is so important. I'm not a Luddite, and I do think generative AI will have huge benefits. I just think it needs to be built in the right way.

Is it too late for companies to repair trust with creators?

I don't think it's too late at all. The thing is, we're still very early in generative AI. You look at where we are at with music generation models. Google released that amazing Lyria model yesterday, and the models are going to keep getting better. But we are still early. Yes, ChatGPT had a hundred million users very quickly, but what are their retention figures? I don't know. This technology hasn't taken over the world yet.

So I think there's a lot of room to play, and ultimately people are working on the next generation of models right now. The models that are live today will not be live in four months time, let alone a year's time. And so as people are building these models, they get a chance to choose how they're building those, and what data they're training on. And I think creators and rights holders are open to the conversation.

There's an ongoing lawsuit against Stability AI that's I think brought by illustrators and artists, and it has to do with the inputs, the copyrighted work that was used to train the models. I'm curious about your thoughts on that and whether some of the arguments that they're making regarding fair use apply to music as well.

I would rather not comment on lawsuits and stuff, if that's okay.

Okay. If you can't answer this either, we can skip, but Casey recently interviewed Rebecca Tushnet, a professor at Harvard Law, who said her view is that a lot of AI output should be un-copyrightable because it doesn't reflect human authorship.

So I guess this is the flip side: ‘is the output copyrightable?’. Look, she's much more knowledgeable than I am, so I trust her more than I would trust me on this. I think there are arguments on both sides, whether output should be copyrightable. I think it's relatively likely that copyright as an idea is going to be less applicable to AI-generated output, just because it clearly operates very, very differently. It is firstly happening at a vast, vast scale, and will continue to do so, often with minimal human intervention. So yeah, I think there's probably something to what she says, but yeah, honestly, it's not my area of expertise.

Okay. This is my last spicy Stability AI question, I promise. But Stability AI CEO Emad Mostaque went on Hard Fork a while back and Casey asked him a question about artists being fearful, annoyed, even angry that their work is being used to train AI models. And he said it comes down to the intention of the individual user, the person using the tools. He differentiated between someone who intentionally is breaking copyright law and someone who does something unethical by asking the model to mimic the styles of existing artists. But he specifically said the company's tools "would never recreate an image" because it's not a compression algorithm. I'm curious how that applies to music, and if you agree with his thinking there.

The first thing I'm going to say is Emad is very thoughtful about these things, and I have a lot of respect for him and what he's done. Honestly, the question of can these systems recreate the inputs? I don't know what the answer to that is. Can they? Given they're all stochastic processes, presumably it's at least possible, like monkeys and typewriters. But I think that's kind of irrelevant. I guess the question ultimately comes down to who is responsible if there is a copyright infringement.

I think, again, it kind of comes back to this point of what is a fair use. And actually I think the problem with copyright and fair use is that it's there to be interpreted. Fair use itself is not clearly defined. Everyone comes up with their own interpretation, honestly. And AI companies will come up with one interpretation that suits them, and content companies and rights holders will come up with another interpretation. So frankly, what you see is you see a lot of extreme views on Twitter arguing.

The reality is almost certainly somewhere in the middle. I think content holders need to understand that generative AI companies believe they're doing a really good thing. And in many ways, I think there are going to be huge benefits of these models, and this isn't just some evil corporation trying to steal their stuff. But generative AI companies need to be really mindful that these are creators’ livelihoods, and they've been brought up as musicians or writers or whatever being told about the value of copyright and told that if they just write this stuff or write this song or write this book for free, then they're hoping that one day they'll get some money from it because they own the copyright and that copyright is really what they rely on.

And so I think generative AI companies need to respect that. So I think we just need more understanding from both sides.

Can you sketch out a vision of how AI music would work and what the ideal approach is, in your mind?

We built one of these at Stability. We built a system and a model that we're really proud of. As an artist, as a creator, I think you need the ability to say, "No, I don't want my stuff to be trained on." You need to be remunerated in some way — if you want remuneration. (Not all creators want remuneration from these models.) And ultimately, I think you probably need, and this is maybe one of the trickier aspects, but it's something that companies like BRIA are doing well on the image side, you need to figure out some compensation model that pays the people who created the content.

Obviously if I'm going onto YouTube Shorts and creating a song in the style of T-Pain, it's pretty clear that T-Pain will be getting more money than all the other musicians whose music was used to train on. But then you can also do that to the base model layer.

It is not a solved problem how you do that at the moment, and more research is needed, but I think what companies like BRIA are doing is they're showing that through combining a technical solution that kind of works with a business solution that all the rights holders are happy with, then you get to something that works for everyone. And that's the key thing. I think you need to reach a model where all parties are happy with it, sign up to it, accept it, and it's why consent is the be all and end all.

It's interesting to think about how this will impact creativity. If you're incentivized to create music that makes it easier for other people to create music off of using generative AI, it could help the rich get richer. The Ed Sheerans of the world will continue to make more than the Caroline Shaws of the world, because that music is more accessible.

I mean, I think it's pretty clear that the accessible music makers are going to keep making the most money. You're speaking to a choral composer. I write music for choir.

There are people who are paid more for their music than I am. Let's put it that way.

Talk about this edition with us in Discord: This link will get you in for the next week.

OpenAI

Negotiations continue between the pro-Altman faction and the board over his potential return. Here’s some of what we’ve been reading over the past day.

Before Altman’s ouster, there were long-simmering tensions on the board — including Altman recently criticizing board member Helen Toner about a research appear that he feared put OpenAI in a negative light relative to its rival Anthropic. (Cade Metz, Tripp Mickle and Mike Isaac / New York Times)
OpenAI’s board of directors reportedly approached Anthropic co-founder and CEO Daniel Amodei with a merger proposal after Sam Altman was fired, in an effort to persuade Amodei to replace Altman. (Stephanie Palazzolo / The Information)
747 out of 770 – about 95 percent – of employees at OpenAI reportedly signed a letter threatening to quit and join Microsoft if the board doesn’t reinstate Altman. (George Hammond and Madhumita Murgia / Financial Times)
Altman has employees on his side – and some are attributing the support to his charisma, ability to inspire, and how he shapes narratives. (Madhumita Murgia and George Hammond / Financial Times)
Sam Altman and interim CEO Emmett Shear, along with at least one other board member and some investors, are reportedly in discussions about Altman possibly being reinstated. (Edward Ludlow, Emily Chang and Ashlee Vance / Bloomberg)
Former president Greg Brockman, who quit after Altman was fired, is still posting updates about ChatGPT though, including one about voice narration. (Kyle Wiggers/ TechCrunch)

Governing

Epic v. Google: Don Harrison, Google’s head of global partnerships, confirmed that the company made a deal with Spotify for Android payments, where Spotify only paid 4 percent as opposed to the usual 15 percent for those using Google’s payment processor. (Adi Robertson and Sean Hollister / The Verge)
- Epic CEO Tim Sweeney said Google’s Android operating system is a “fake open platform” that’s actually closed to developers. (Malathi Nayak / Bloomberg)
President Biden and Vice President Harris both now have Threads accounts, ahead of the 2024 elections. Great, now delete those X accounts. (Maria Curi / Axios)
A Senate panel subpoenaed the CEOs of X, Discord, and Snap to testify at a hearing about online child safety, after “repeated refusals” from the companies to cooperate. (Cristiano Lima / Washington Post)
Elon Musk is suing Media Matters, an organization that monitors online misinformation, for defamation, claiming that it misrepresented the amount of antisemitic content on X. (Sharon Knolle and Mike Roe / The Wrap)
- At an all-hands meeting led by Yaccarino, the CEO reportedly said that customers have paused investments because of “misleading or manipulated information”, and staff should “by all means, put your heads together to bring new revenue”. (Kylie Robison / Fortune)
- But the defamation suit seems to confirm what it claims is defamatory – that ads from major companies were in fact seen next to antisemitic content. (Devin Coldewey / TechCrunch)
- Texas attorney general Ken Paxton also opened an investigation into Media Matters for “potential fraudulent activity”, after Musk suggested conservative attorneys general look into fraud charges. A terrible assault on speech rights from a uniquely awful public official. (Adi Robertson / The Verge)
- The NFL, one of X’s major advertisers, denounced hate speech but stopped short of saying it was suspending ads on the platform. (Oliver Darcy / CNN Business)
X’s Community Notes is failing to combat a surge in misinformation on the platform relating to the Israel-Hamas conflict, with the feature working slowly and inconsistently, according to a Bloomberg analysis. (Davey Alba, Denise Lu, Leon Yin and Eric Fan / Bloomberg)
The Trump Media and Technology Group, parent of Truth Social, is suing 20 news outlets over reports that the platform had lost $73 million since launch, saying that the reports were “an utter fabrication” and an “attack”. (Robert Burnson / Bloomberg)
Binance CEO Changpeng Zhao pled guilty to criminal charges for anti-money laundering and US sanctions violations, including allowing transactions with terrorist groups like Hamas. Binance will pay more than $4 billion in penalties. (Chris Strohm, Matt Day, Ava Benny-Morrison and Allyson Versprille / Bloomberg)
The FTC approved a new streamlined process for investigations related to AI, making it easier for staff to issue demands for documents. (Reuters)
Netflix and Amazon are increasingly backing away from films that offer commentary on the Modi administration in India, so as not to offend the government and affect their thriving businesses in the country. (Gerry Shih and Anant Gupta / Washington

Industry

Anthropic released Claude 2.1, an improved version of its AI assistant that has better context length, accuracy, and integration capabilities. Even more impressively, the company also has a CEO. (Kristi Hines / Search Engine Journal)
Stability.AI announced Stable Video Diffusion, its first foundation model for generative AI video based on its Stable Diffusion image model. (Stability.AI)
Some Firefox and Edge users are reportedly experiencing a delay in loading YouTube videos on the browser, which goes away when they switch to Chrome. YouTube says the delay is related to ad blockers, not browsers. (Aamir Siddiqui / Android Authority)
Meta released the Meta Content Library and API, which lets researchers access publicly available data on Facebook and Instagram. I got a demo of this and it’s super cool. Credit to Meta for putting together a really nice API for researchers. (Tate Ryan-Mosley / MIT Tech Review)
Twitter’s former head of trust and safety Del Harvey talks about content moderation and how generative AI is making moderation more difficult. (Lauren Goode / WIRED)
Amazon and the NFL are betting that the first “Black Friday” game, set to stream on Prime Video, is going to combine football and holiday shopping to boost ratings and sales. (Joe Flint and Sebastian Herrera / The Wall Street Journal)
Zoom reported Q3 revenue that beat analyst expectations, with strong enterprise sales and a revenue increase of more than 3 percent, to about $1.14 billion. (Brody Ford / Bloomberg)

Those good posts

Please send us good posts, we’re dying.

(Link)

Talk to us

Send us tips, comments, questions, and ethically sourced training data: casey@platformer.news and zoe@platformer.news.

OpenAI's alignment problem

Casey Newton — Tue, 21 Nov 2023 01:47:51 GMT

Less than two months ago, on stage at the Code Conference, I asked Helen Toner how she thought about the awesome power that she’d been entrusted with as a board member at OpenAI. Toner has the power under the company’s charter to halt OpenAI’s efforts to build an artificial general intelligence. If the circumstances presented themselves, would she really stop the company’s work and redirect employees to working on other projects?

At the time, Toner demurred. I had worded my question inelegantly, suggesting that she might be able to shut down the company entirely. The moment passed, and I never got my answer — until this weekend, when the board Toner serves on effectively ended OpenAI as we know it. (She declined to comment when I emailed her.)

By now I assume you have caught up on the seismic events of the past three days at OpenAI: the shock firing on Friday of CEO Sam Altman, followed by company president Greg Brockman quitting in solidarity; a weekend spent negotiating their possible returns; ex-Twitch CEO Emmett Shear being installed by the board as OpenAI's new interim CEO; and minority investor Microsoft swooping in to create a new advanced research division for Altman and Brockman to run.

By mid-afternoon Monday, more than 95 percent of OpenAI employees had signed a letter threatening to quit unless Altman and Brockman are reinstated. It seems all but certain that there will be more twists to come.

I found this turn of events as stunning as anyone, not least because I had just interviewed Altman on Wednesday for Hard Fork. I had run into him a few days before OpenAI’s developer conference, and he suggested that we have a conversation about AI’s long-term future. We set it up for last week, and my co-host Kevin Roose and I asked about everything that has been on our minds lately: about copyright, about open-source development, about building AI responsibly and avoiding worst-case scenarios. (We’ve just posted that interview, along with a transcript, here.)

The days that follow revealed fundamental tensions within OpenAI about its pace of development and the many commitments of its CEO. But in one fundamental respect, the story remains as confusing today as it did on Friday when the board made its bombshell announcement: why, exactly, did OpenAI’s board fire Sam Altman?

The official explanations have proliferated. The board’s original stated reason was that Altman “was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” In an all-hands meeting the next day, the company’s chief scientist and a member of the board, Ilya Sutskever, suggested that the removal had been necessary “to make sure that OpenAI builds AGI that benefits all of humanity.” To some employees, the remark suggested that the firing may have been connected to concerns that OpenAI was unduly accelerating the development of the technology.

Later on Saturday, the chief operating officer, Brad Lightcap, ruled out many possible explanations while blaming the situation on poor communication. “We can say definitively that the board’s decision was not made in response to malfeasance or anything related to our financial, business, safety or security/privacy practices,” he told employees. “This was a breakdown in communication between Sam and the board.”

The following evening, after he was named CEO, Shear also ruled out a safety explanation. “The board did *not* remove Sam over any specific disagreement on safety, their reasoning was completely different from that,” he posted on X, without elaborating.

The cumulative effect of these statements was to all but exonerate Altman. In a few short years, he had led the company from a nonprofit research effort to a company worth as much as $90 billion to investors. And now he was being driven out over some unspecified miscommunications? It didn’t add up.

In their silence, the board ensured that Altman became the protagonist and hero of this story. Altman’s strategic X posts, cleverly coordinated with his many allies at the company, gave him the appearance of a deposed elected official about to be swept back into power by the sheer force of his popularity.

By the time Susteveker apologized for his role in the coup Monday morning, and expressed his desire to reinstate Altman and Brockman to their roles, I had officially lost the plot. By some accounts, Susteveker had spearheaded the effort to remove Altman from his post. Now that he’s had some time to read the room, apparently, he’s changed his mind.

So far much of the attention this story has received has focused, understandably, on the value that has been destroyed. ChatGPT is the most compelling product in a generation, and had it been left to develop according to plan it likely would have cemented OpenAI’s position as one of the three or four most important technology companies in the world.

But to most people, it will never matter how much OpenAI was worth. What matters is what it built, and how it deployed it. What jobs it destroyed, and what jobs it created. What capabilities the company’s technology has — and what features it disabled before releasing.

Navigating those tensions is the role of OpenAI’s nonprofit board. And while it botched its role badly, that can’t be the end of the story.

Someone has to navigate those tensions. And before we blame this weekend’s fiasco on OpenAI’s unusual structure, or on nonprofit governance broadly, it’s worth considering why the board was set up this way in the first place.

II.

When they launched OpenAI in 2015, the founders considered several models. One was a public-sector project funded by the government — but our government being what it is, they saw no feasible way to spin up such an effort. Another was a venture-backed startup. But the founders believed the superintelligence they were trying to build should not be concentrated in the hands of a single for-profit company.

That left the non-profit model, which they hoped “would be the most effective vehicle to direct the development of safe and broadly beneficial AGI while remaining unencumbered by profit incentives." They set out to raise $1 billion — Elon Musk said he contributed $100 million — and set to work.

Four years later, they discovered what everyone else who tries to train a large language model eventually realizes: even $1 billion doesn’t get you very far. To continue building, they would have to take money from private investors — which meant setting up a for-profit entity underneath the nonprofit, similar to the way the Mozilla Foundation owns the corporation that oversees revenue operations for the Firefox browser, or how the nonprofit Signal Foundation owns the LLC that operates the messaging app.

While the membership of the board has changed over the years, its makeup has always reflected a mix of public-benefit and corporate interests. At the time of Altman’s firing, though, it had arguably skewed away from the latter. In March, LinkedIn co-founder Reid Hoffman left the board; he co-founded a for-private rival two months later.

That left three OpenAI employees on the board: Altman, Brockman, and Sutskever. And it left three independent directors: Toner, Quora co-founder Adam D’Angelo, and entrepreneur Tasha McCauley. Toner and McCauley have worked in the effective altruism movement, which seeks to maximize the leverage on philanthropic dollars to do the most good possible.

The reputation of effective altruism cratered last year along with the fortunes of one of its most famous adherents, Sam Bankman-Fried. Bankman-Fried had sought to generate wealth as rapidly as possible so he could begin giving it all away, ultimately defrauding FTX customers out of billions of dollars. Taken to this extreme, EA can create harmful incentives.

But for rank-and-file EAs, the movement’s core idea was to apply some intellectual rigor to philanthropy, which too often serves only to flatter the egos of its donors. EA groups often invest in areas that other funders neglect, and were early funders of research into the potential long-term consequences of AI. They feared that advances in AI would begin to accelerate at a rapid clip, and then compound, delivering a superintelligence into the world before we made the necessary preparations.

The EAs were interested in this long before large language models were widely available. And the thing is, they were right: we are currently living at a time of exponential improvement in AI systems, as anyone who used both GPT-3 and GPT-4 can attest. AI safety researchers will be the first to tell you that this exponential progress could hit a curve: that we won’t be able to solve the research questions necessary to see a similar step-change in functionality if and when a theoretical GPT-5 is ever trained and released.

But given the progress that has been made just in the past couple years, the EAs would say, shouldn’t we do some worst-case-scenario planning? Shouldn’t we move cautiously as we build systems that aid in the discovery of novel bioweapons, or can plan and execute schemes on behalf of cybercriminals, or who can corrupt our information sphere with synthetic media and hyper-personalized propaganda?

If you are running a for-profit company, questions like these can be extremely annoying: they can suffocate product development under layers of product, policy, and regulatory review. If you’re a nonprofit organization, though, these can feel like the only questions to ask. OpenAI was founded, after all, as a research project. It was never meant to compete on speed.

And yet, as The Atlantic reported over the weekend, ChatGPT itself launched last year less out of the company’s certainty that it would be beneficial to society than the fear that Anthropic was about to launch a chatbot of its own. Time and again over the next year, OpenAI’s for-profit arm would make moves to extend its product lead over its rivals. Most recently, at its developer conference, the company announced GPTs: custom chatbots that represent a first step toward agents that can perform higher-level coordination tasks. Agents are a top concern of AI safety researchers — and their release reportedly infuriated Sutskever.

And during all this time, was Altman squarely focused on OpenAI? Well … not really. He was launching Worldcoin, his eyeball-scanning crypto orb project. He was raising a new venture fund that would focus on “hard” tech, according to Semafor. He was also seeking billions to create a new company that would build AI chips, Bloomberg reported.

Silicon Valley has a long history of founders running multiple companies. Steve Jobs had roles at Apple and Pixar simultaneously; Jack Dorsey served as CEO of Twitter and Square at the same time.

What’s different in the Altman case is that OpenAI was driven by a public mission, and one that would seem to foreclose certain for-profit extracurricular work. If OpenAI is designed to promote cautious AI development, and its CEO is working to build a for-profit chip company that might accelerate AI development significantly, the conflict seems obvious. And that’s before you even get into the question of what Altman told the board about any of this, and when.

None of which is to say that the board couldn’t have found a way to resolve this situation without firing Altman. Particularly given the billions of dollars at stake. But the board’s job is explicitly to ignore any concerns about money in favor of safe AI development. And they would not be the first people to break with Altman over fears that OpenAI has not been true to its mission.

III.

However valid these concerns may have been, it’s now clear that the board overplayed its hand. The success of ChatGPT had made the nonprofit an afterthought in the minds of its 700-plus employees, to say nothing of the world at large. And Altman is a popular leader, both inside OpenAI and as a kind of roving AI diplomat. The board was never going to win a fight with him, even if it had communicated its position effectively.

Now the board stands to lose everything. One former employee described OpenAI to me over the weekend as “a money incinerator.” As successful as it is, ChatGPT is not close to being profitable. There’s a reason Altman was out raising new capital — the company needs it to keep the lights on. And now, with everything that has transpired, it is extremely difficult to imagine how the company could raise the funds necessary to power its ambitions — assuming it even has any employees left to execute on them.

On one hand, history may show that the board did serve as good stewards of their mission. On the other, though, the board was also responsible for being a good steward of OpenAI as an institution. In that, it failed entirely.

It is understandable to look at this wreckage and think — well, that’s what happens when you mix nonprofit governance with for-profit incentives. And surely there is something to that point of view.

At the same time, though, it’s worth asking whether we would still be so down on OpenAI’s board had Altman been focused solely on the company and its mission. There’s a world where an Altman, content to do one job and do it well, could have managed his board’s concerns while still building OpenAI into the juggernaut that until Friday it seemed destined to be.

That outcome seems preferable to the world we now find ourselves in, where AI safety folks have been made to look like laughingstocks, tech giants are building superintelligence with a profit motive, and social media flattens and polarizes the debate into warring fandoms. OpenAI’s board got almost everything wrong, but they were right to worry about the terms on which we build the future, and I suspect it will now be a long time before anyone else in this industry attempts anything other than the path of least resistance.

Correction, 11/21: I updated the language in this story to reflect that while they set out to raise $1 billion, it’s unclear that the OpenAI team actually reached that figure.

On the emergency podcast today: Kevin and I trade notes on the weekend’s events. And then, we present our interview from Wednesday with Altman, in the final hours before OpenAI got turned upside-down. (This episode will take the place of our usual Friday episode; we’ll be back next week after the holiday.)

Governing

Google v. DOJ: As the case came to a close, Judge Amit Mehta was undecided, saying had “no idea” what he was going to do. Closing arguments somehow don’t take place until May. (Reuters)
A federal judge declined to let Elon Musk avoid a deposition in an FTC investigation and rejected X’s attempt to overturn an FTC order that imposed requirements on safeguarding personal user data. (Joseph Menn / Washington Post)
Apple paused all advertising on X after its ads were placed along far-right content and Musk backed an antisemitic post. Good! (Ina Fried / Axios)
- Disney, Warner Bros Discovery, Paramount, and Lions Gate have also paused ad campaigns on X after Musk’s endorsement of an antisemitic post. (Jonathan Vanian / CNBC)
- The European Commission also decided to temporarily suspend advertising on X, citing “widespread concerns relating to the spread of disinformation”. (Nicholas Vinocur / POLITICO)
- But right-wing media companies and influencers are pledging support for Musk, saying they will advertise on X to help recoup the losses in ad revenue. (Matt Binder / Mashable)
Microsoft blocked the term “Disney” from its AI image generator tool after a viral trend saw users creating Disney film posters of their dogs, raising copyright concerns. (Cristina Criddle / Financial Times)
Apple filed a challenge against the European Digital Markets Act in an effort to avoid the App Store being designated as gatekeeper. If the App Store isn’t a gatekeeper, what is? (Martin Coulter / Reuters)
The Italian Competition Authority is probing Meta over how it marks branded content and how the tool is used in relation to promotional content on Instagram. (Giulia Petroni / The Wall Street Journal)
France, Germany and Italy have reportedly reached an agreement on AI regulation, with an emphasis on regulating how AI systems are applied rather than limiting the technology itself. (Andrea Rinke / Reuters)

Industry

X’s job search tool is now live on the web version of the app, with open positions across a number of tech companies including Musk’s Space X, Neuralink, Tesla, and x.ai. I wonder why there are so many openings at those companies. (Karissa Bell / Engadget)
The updated developer terms on X suggests that EU researchers will once again have access to the platform’s API. But likely only due to European regulation. (Ivan Mehta / TechCrunch)
Amazon is reportedly set to cut several hundred roles in the Alexa division, as it shifts its focus to generative AI. (Todd Bishop / GeekWire)
Meta AI researchers said its image generation model, Emu, had made breakthroughs in text-to-image editing and video generation. (Mike Wheatley / Silicon Angle)
Meta has reportedly split up its team tasked with understanding and preventing AI-related harms and diverting employees to its generative AI team instead. (Kalley Huang and Sylvia Varnham O’Regan / The Information)
Ex-Meta employees are forming their own startups and venture capital firms selling trust and safety technology amid the Israel-Hamas conflict — and finding a robust market. (Jonathan Vanian / CNBC)
Google is reportedly delaying the release of its next-generation Gemini AI model till the first quarter of next year for cloud customers and business partners. (Jon Victor / The Information)
Google announced plans to gradually phase out third-party cookies in Chrome, with a 1 percent user testing period beginning early next year. (Mayank Parmar / Bleeping Computer)
Snap CEO Evan Spiegel is putting pressure on his staff to work harder, apparently walking around the company’s headquarters and noting who has been working late in the office. (Erin Woo / The Information)
Tinder is revamping its profile pages with a bunch of features, including profile prompts, quizzes, basic info tags, and dark mode. (Ivan Mehta / TechCrunch)
Discord is shutting down its experimental AI chatbot Clyde and deactivating it at the end of the month. (Tom Warren / The Verge)

Those good posts

Send us good posts from Bluesky and Threads! Just reply to this email.

(Link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

Notion gets thrown for a Loop

Casey Newton — Fri, 17 Nov 2023 01:23:43 GMT

Today let’s talk about one of my favorite productivity companies and the challenge it now faces from a tech giant. It’s a story that speaks to the challenges of competing with software bundles in a world with only minimal antitrust enforcement — even when the challenger’s product is significantly better.

Seven years ago this month, Slack was riding high. One of the fastest-growing enterprise software companies of all time, it hit a $100 million run rate and was valued at $3.8 billion. The company arrived in an era when newly smartphone-savvy workers were choosing their own tools to bring to work, and for a moment it looked as if it might be a generation-defining company.

Then Microsoft entered the picture. Teams, a straightforward clone of Slack’s workplace communication app, didn’t arrive with a bold new design or fresh set of features. But while it looked to fans of Slack like a pale imitation, Teams boasted two crucial advantages. One, it’s part of the Microsoft 365 bundle. That meant any workplace that already paid for Word, Excel, and PowerPoint could now get Slack for no extra charge. And two, Teams is backed by Microsoft’s formidable sales force, which helps the company sell into the large enterprises that are traditionally most profitable for software services.

In a move that the company came to regret more or less instantly, Slack “welcomed” Teams to the world with a smarmy full-page newspaper ad. In retrospect, it was the moment that the company peaked. Over the next four years, its user base would triple to about 12 million people. Over that same time period, Microsoft took Teams from zero to 115 million users. Slack went public in 2019, but did not turn a profit this year, and after its stock floundered the company sold to Salesforce for the (admittedly incredible) price of $27.7 billion.

Since then, despite a recent redesign, Slack has more or less been treading water. The division just named its third CEO in a year. A fun party game you can play in San Francisco is to try to find a Salesforce employee who thinks that buying Slack was a good strategic move for their company. (I’ve never succeeded.)

Slack made more than its share of mistakes along the way, including on the product side. Even if it had executed perfectly, though, there’s a good chance the outcome here would have looked more or less the same.

I thought about all this today when reading about Loop. Like Teams, it’s a new app from Microsoft, and also like Teams, it’s a clone of a successful product from a challenger. Loop is a clone of Notion, a 7-year-old workplace collaboration app that boasts more than 20 million users.

As of today, Loop is now included in the Microsoft 365 bundle. Here’s Jay Peters at The Verge:

Loop lets you use flexible, collaborative workspaces and pages to make it easier to cooperate on work. If you’re familiar with Notion’s interface at all, Loop looks and feels remarkably similar — right down to the ability to easily access a bunch of tools and formatting options by typing the forward slash key (which pulls up what Microsoft calls the “insert menu”).
But because Loop is built by Microsoft, that means it has some useful integrations with other Microsoft software. For example, you can take parts of Loop pages and share them across other Microsoft apps like Teams chat and Outlook, which seems like a handy way to be able to work on things together without being forced to context switch between apps. Microsoft’s AI-powered Copilot assistant is also available within Loop, which can help with things like drafting text and summarizing pages in the app.

Like Teams before it, there is little in Loop today to entice any Notion user to switch. But as with Teams, Loop isn’t really aimed at Notion users. It’s aimed at people who might have used Notion eventually, if their workplaces had ever bought it for them.

And Microsoft has the same advantages here that it had against Slack. Anyone with a 365 account can now use Loop at no extra charge. And Microsoft’s sales force can now tell thousands of clients that the value of its bundle just increased significantly, as workplaces may no longer have any need to buy Notion. (Notion plans start at $8 per user per month; Microsoft’s basic 365 plan for businesses is actually $2 cheaper.)

Of course, some businesses — particularly those that prefer to run on Macs — will never purchase Microsoft 365. And I imagine small- and medium-sized businesses may continue to prefer the lovingly crafted, pixel-polished Notion, particularly in the more design-forward enclaves of Silicon Valley.

But assuming Loop scoops up tens of millions of users in the same way Teams did, that leaves Notion with far fewer businesses to sell into — and a much more challenging path to realize the $10 billion valuation investors gave it this year.

None of this will come as any surprise to executives at Notion. Loop was announced in 2021, and since then Notion has continued to roll out new features at an admirable pace. (One that far exceeded Slack’s innovations in the year after Teams was announced, for what it’s worth). Over the past year, the company has invested heavily in features that make use of artificial intelligence: for an extra monthly fee, Notion will write documents for you, summarize them, extract key information from links stored in databases, and much more.

This week, Notion launched a feature I had asked for back in August: a tool that uses generative AI to let you query your entire Notion instance. Many companies use Notion to create internal wikis, and Q&A offers a nice way to find onboarding documents, human resources information, and other important but typically hard-to-find internal information. (David Pierce has a good write-up of Notion Q&A in The Verge.)

For the past few weeks, I’ve been using Q&A to essentially chat with the database of links that I store in Notion to power the links section of the newsletter. The first couple weeks, when Notion was using a less capable model, were relatively rough. Answers were inconsistent and sometimes wrong. Over the past few days, it’s gotten much better, though it’s still very much in beta and more of a curiosity than a core part of my workflow.

AI seems like absolutely the right focus for Notion as it works to differentiate itself from competitors. But how differentiated can a company be in AI when its competition is Microsoft? At its Ignite conference this week, Microsoft announced its own AI chatbot strategy, and its AI Copilot comes standard in Loop.

On one hand, Loop’s existence is a testament to the power of competition. Notion’s highly flexible workspaces helped to reinvent the idea of what a digital document could be, and in the process jolted Microsoft and other rivals out of their slumber.

The question, now that the giants have awakened, is whether Notion can continue to compete in any meaningful sense of the word. The company has wisely kept any opinions about Loop to itself, avoiding Slack’s mistake of focusing attention on a rival. But the fundamental dynamics that doomed Slack still seem to apply.

After it went public, and as its business faltered, Slack filed a complaint with the European Commission accusing Microsoft of unfairly stifling competition in the marketplace by bundling Teams. This July — three years after Slack sounded the alarm — the commission opened a formal probe into the matter.

“The Commission is concerned that Microsoft may be abusing and defending its market position in productivity software by restricting competition in the European Economic Area … for communication and collaboration products,” the commission said in announcing the probe. “These practices may constitute anti-competitive tying or bundling and prevent suppliers of other communication and collaboration tools from competing, to the detriment of customers.”

If the commission makes a formal finding that those statements are true — and Lord knows how long that might take — it seems to me that what applies to Slack might apply to Notion as well.

In the meantime, Notion’s advantage may be that it cares about the product in a way that is existential for the company. But now that a cheaper, bundled alternative is available, history suggests the company will struggle to get others to care nearly that much.

On the podcast this week: I tell Kevin about my field trip to Humane’s office. Then, we debate YouTube’s new deepfake policies. And finally, SciFi Foods CEO Joshua Mark swings by to serve us a lab-grown Thanksgiving.

Governing

Elon Musk agreed with a post on X that claims Jewish communities push “hatred on whites” and that referenced “hordes of minorities” flooding Western countries. (David Goldman / CNN Business)
Epic v. Google: Google CEO Sundar Pichai testified in this trial soon after he testified in the DOJ antitrust trial, confirming the huge percentage of Safari search revenue that Google shares with Apple. (Nico Grant / The New York Times)
Meta will allow political ads that question the legitimacy of the 2020 elections on Facebook and Instagram, ahead of the 2024 elections. A terrible step backwards for the company. (Salvador Rodriguez / The Wall Street Journal)
Meta says app stores should be responsible for age-gating. The company is calling for legislation to require parental approval when underage users download apps. (Cristiano Lima and Naomi Nix / Washington Post)
Cartel-backed smugglers are exploiting Snapchat, TikTok and other apps to find drivers to transport migrants across the US-Mexico border. (Julia Love / Bloomberg)
Amazon is removing seven brands of eye drops from its site after the FDA sent a letter to CEO Andy Jassy saying the products were not recognized as safe and effective. (Amanda Holpuch / The New York Times)
TikTok is challenging its categorization as a “gatekeeper” under the European Digital Markets Act and Meta is appealing the decision to designate Facebook Messenger and Marketplace as “core” services. (Javier Espinoza / Financial Times)
- Meta also challenged its “gatekeeper” status. (Supantha Mukherjee / Reuters)
The UK’s minister for AI and intellectual property, Viscount Jonathan Camrose, said there will be no regulation on AI in the short term, citing concerns about curbing industry growth. This viscount loves capitalism! (Daria Mosolova / Financial Times)
Amazon is reportedly funding advocacy groups lobbying against Microsoft’s efforts to become a major cloud computing contractor for governments. (Emily Birnbaum / Bloomberg)
YouTube’s updated guidelines will allow monetization on videos that display nudity while breastfeeding and “non-sexually graphic dancing”. (Aisha Malik / TechCrunch)
Facebook and YouTube are the most popular sites for news in the US, a new report by the Pew Research Center found. (Sarah Scire / Nieman Lab)
An email server flaw was used to steal data from governments in Greece, Moldova, Tunisia, Vietnam, and Pakistan, Google’s Threat Analysis Group found. (Katie Malone / Engadget)
AI companies are inadvertently using underage labor in their global data supply chain, often exposing young teens to traumatic content. (Niamh Rowe / WIRED)
An analysis of AI tools by Common Sense Media warns parents that Snapchat’s My AI chatbot is willing to chat with young users about sex and alcohol, and that it misrepresented Snap’s targeted advertising. (Khari Johnson / WIRED)
Social media users, armed with AI image recognition tools, are helping morgues and authorities who are overwhelmed and lack resources to identify otherwise unidentified bodies. (Deidre Olsen / WIRED)
Facial recognition software is often used by police in executing search warrants and arrests, despite its limitations and biases. In one case, it’s led police astray, wrongfully arresting someone despite contradicting evidence. (Eyal Press / The New Yorker)
Ireland’s data protection commissioner and the EU’s top privacy watchdog, Helen Dixon, is stepping down. (Stephanie Bodoni / Bloomberg)
Argentina’s presidential election is becoming a testing ground for AI, with both candidates using AI to create images and videos to promote themselves and attack each other. (Jack Nicas and Lucía Cholakian Herrera / The New York Times)
Japan’s new tax code proposal will make app store operators like Apple and Google pay consumption taxes on content sold by foreign developers. (Ko Fujioka / Nikkei Asia)

Industry

TikTok said it was removing videos promoting Osama bin Laden’s “Letter to America”, which sought to justify the 9/11 attacks. A huge discussion broke out online about how viral the videos were; it seems like most of the attention the TikToks got was on X. (Ted Johnson / Deadline)
Tiktok executives were recently confronted by a number of Jewish celebrities and creators, urging them to address the increase in antisemitism and harassment on the platform. (Sapna Maheshwari / The New York Times)
ByteDance is reportedly testing a feature that lets some creators add paywalls on parts of their videos on TikTok’s sister app in China, Douyin. (Josh Ye / Reuters)
OpenAI temporarily paused new subscriptions to ChatGPT Plus, with CEO Sam Altman saying usage exceeded the company’s capacity following its recent DevDay conference. (Kristi Hines / Search Engine Journal)
Ed Newton-Rex resigned from his role as vice president of Audio at Stability AI, saying he disagrees with the company’s stance that training generative AI models on copyrighted works constitutes “fair use.” (Ed Newton-Rex / Music Business Worldwide)
Microsoft announced several new AI products at Ignite, including new cybersecurity solutions, strengthening infrastructure, and expanded Copilot capabilities for sales and services. (Matt Marshall / VentureBeat)
Bing is now using GPT-4 to generate captions for some of its search results to make them “more relevant and informative”. The captions will be labeled as AI-generated. (Barry Schwartz / Search Engine Land)
Lots of Google product updates today, starting with: Google Photos is rolling out three new features, including auto-grouping Photo Stacks and better categorization of pictures and documents. (Abner Li / 9to5Google)
- Google’s new Titan security key models lets users store passkeys and works with third-party services. (Abner Li / 9to5Google)
- Bard will soon be available to teens with a number of safety features, including a double-check response feature and an AI Literacy Guide. (Abner Li / 9to5Google)
- Search Generative Experience, Google’s AI-powered search, is allowing shoppers to generate gift ideas and items to help with online shopping. (Jess Weatherbed / The Verge)
- Google News is removing support for paid magazines, both for new subscriptions and existing ones. (Ben Schoon / 9to5Google)
- Maps is getting improved transit directions, emoji reactions, and collaborative lists. (Andrew J. Hawkins / The Verge)
- Users can choose to annotate search results using Google’s new “Notes” experiment. Sure to be a vibrant new battleground for content moderators. (Jay Peters / The Verge)
YouTube is expanding its 1090p HD option for premium users to more devices, on Android, Web, and smart TVs. (Aisha Malik / TechCrunch)
YouTube Shorts creators are getting a new generative AI tool called Dream Track, which can generate songs in the style of nine different artists, including Charlie Puth, Demi Lovato, Sia, and T-Pain. (Will Knight / WIRED)
WhatsApp Channels is adding a new feature, stickers, as it crosses the 500 million monthly active user mark. (Tech Desk / The Indian Express)
Instagram is adding new filters, improving editing functions on Reels, and testing the ability to let users create custom stickers from videos and images in their camera roll. (Chris Welch / The Verge)
Threads is now testing the ability for people to use hashtags to tag topics, but without displaying the “#” symbol. (Sarah Perez / TechCrunch)
Meta is partnering with MediaTek to develop custom silicon to improve its AR smart glasses capabilities. (Andrew Romero / 9to5Google)
Apple is extending the free usage period for Emergency SOS satellite features for iPhone 14 users by another two years. (Benjamin Mayo / 9to5Mac)
iPhone will reportedly have support for Android’s messaging standard, RCS messaging, which means users can soon stop fighting about the green and blue bubbles. (Emma Roth / The Verge)
Proton Mail’s new Key Transparency feature uses blockchain technology to allow users to verify email addresses, and CEO Andy Yen wants to be clear: it’s not “some sketchy cryptocurrency”. (Leo Schwartz / Fortune)
Signal president Meredith Whittaker says running the platform will soon cost $50 million as she calls out competitors with ad-based business models. (Andy Greenberg / WIRED)
People are more likely to think pictures of white faces generated by AI are human than photographs of actual people, a team of researchers found. The results do not hold up for images of people of color. (Nicola Davis / The Guardian)
Artist and musician Holly Herndon, along with others in the creative field, use AI to help create art. But that comes with challenges around copyright, consent and compensation for other artists. (Anna Wiener / The New Yorker)

Those good posts

In light of the ongoing, disgusting antisemitism on the platform, we will no longer be including funny X posts in this section or on Casey’s Instagram. There are still lots of funny people on X, but it’s time for them — and us — to move on. (Casey and the Platformer account, which for the past year have used X only to post newsletter and podcast links, will no longer be doing that, either.)

We know this part of the newsletter is a favorite for many of you. It’s one of our favorite parts of putting it together. And while any day over the past year would have been a good day to stop using X, today was our day.

Going forward, we’ll experiment with putting other stuff here. Threads continues to be a great place to discuss news, but it’s not a very funny one. We hope Weird Twitter will migrate there over time. Bluesky is funnier, but we spend less time there. Mastodon is a convention of Linux developers. The landscape is bleak.

We’re committed to finding funny social posts to put here, and we highly encourage you to send stuff our way. So please do. And now, some skeeys we found in today’s Today in Tabs.

(link)

Talk to us

Send us tips, comments, questions, and posts: casey@platformer.news and zoe@platformer.news.

Sponsor a Newsletter

YouTube opens its doors to deepfakes

Casey Newton — Wed, 15 Nov 2023 01:29:25 GMT

Today, let’s talk about platforms’ early moves to moderate the way people create and distribute media created with generative artificial intelligence. Announcements made by YouTube today about its own synthetic media policies suggest that the leverage in moderating deepfakes may not lie where we expected.

I’ll get to the YouTube announcements in a minute. But first, I think it’s helpful to frame how we’ve been approaching digital content moderation up until this point.

In the first era of digital media — from the rise of Facebook onward — we put the primary responsibility for moderating content onto the user and the platform. US users carry the legal liability for most of what they post, thanks to Section 230; similar laws shield platforms from legal responsibility in other big markets. Platforms have some legal responsibilities — they often have to remove terrorist content, for example, or CSAM — but most of the moderation they do serves business interests. Most people don’t like spending time or money in places full of hate speech and other harms, and so platforms remove it.

The AI era of digital media has introduced a third character into the moderation stack: the tool. Generative AI tools can create realistic depictions of human beings, mimic their voices, and animate them on video. They can create erotica of various kinds. They can, if left unchecked, offer detailed instructions on how to build weapons.

We can debate how new this really is. Adobe Photoshop can also create realistic depictions of human beings, and erotica of various kinds. Talented actors can mimic voices and create convincing likenesses on video. You can get pretty far in building a weapon just by Googling.

Still, in the first era of digital media, we saw relatively little pressure on tools like these to perform content moderation during the act of creation. If you draw a naked human form using Photoshop, Adobe won’t interrupt you to ask you what you’re doing. We don’t generally prohibit legal prohibitions on actors from mimicking people. Google won’t delete your account based on your search activity alone.

There are some good reasons for this. One, historically we’ve mostly agreed that what you do on your computer is your own business, as long as it’s not hurting anyone. Two, and maybe more importantly, we’ve been able to count on platforms intervening to stop the spread of harmful material. A deepfake of Joe Biden that I create on my computer for my own enjoyment isn’t going to hurt anyone. That same deepfake, posted on Instagram, might cause chaos. And so we generally put the most pressure to remove material on the platform, rather than the tool.

Then AI came along, and flipped things upside-down.

Take the biggest generative AI platforms: OpenAI’s ChatGPT and DALL-E; Anthropic’s Claude; Google’s Bard; Midjourney; Stability AI. All restrict the use of their tools to produce sexual and violent images, even if those images never leave the user’s computer. Some go further, preventing the tool from creating images of corporate logos or other copyright material. OpenAI banned the use of ChatGPT to create political messages. DALL-E and other AI tools also generally prohibit you from creating images of politicians. In my experience, chatbotsplatforms are also extremely skittish about answering even basic questions about sexual health.

These restrictions reflect a general sense of caution in most AI companies. They saw the beating that social media companies took over the past half-decade, both among lawmakers and in their public reputation, after they under-invested in content moderation. At the same time, most of the AI executives I’ve spoken to are genuinely concerned about the potential for misuse that their tools represent.

That’s a good thing. At the same time, it means that people often can’t use these tools as they used their pre-AI equivalents. It’s a downgrade in usability, and the AI platforms generally do a bad job of educating users about their policies and rehabilitating them if they violate them. (One employee of a big AI company told me that 99 percent of people who got banned from the platform had been trying to create text-based erotica.)

In the old days, digital tools were permissive, and digital platforms were restrictive. Today, digital tools are restrictive. And the platforms?

Well, that brings us to YouTube.

II.

On Tuesday, YouTube announced its “approach to responsible AI innovation.” It is not a list of prohibitions on the ways you can use AI on YouTube. Rather, it the policies serve as a general green light for people to post synthetic media widely on YouTube.

The “responsible” part of the innovation comes in new labels that, at some point “over the coming months,” will be required on synthetic media. In most cases, the label will appear as metadata underneath the video. And “for certain [unspecified] types of content about sensitive topics,” the label will appear overlaid on the video itself.

There are circumstances under which YouTube will remove a manipulated video: if it depicts shocking violence, for example, or if it poses a risk of “egregious harm.”

For the most part, though, you’ll be able to post synthetic media on YouTube as you wish. You can post your deepfake of Joe Biden, or a lesser-known YouTube creator, or a person at your college, and be in compliance with YouTube’s guidelines.

If the subject of your deepfake doesn’t like what you have done with their face and voice, they can fill out a form. The responsibility for addressing harm here falls not on the platform, or the user, or the tool, but on the victim.

And under what circumstances will YouTube respond to your complaint? “We’ll consider a variety of factors when evaluating these requests,” is all that the blog post’s authors, YouTube vice presidents of product management Jennifer Flannery O’Connor and Emily Moxley, have to say about that. (A spokeswoman told me that if you are an average citizen and get deepfaked and don’t like what you see on YouTube, that alone could be enough to get the video removed under the company’s privacy policy. But it all seems a slippery, at least for the moment.)

There are benefits to permitting a wide variety of speech, particularly political speech, and the policy YouTube described today could accomplish that. It’s carving out meaningful permissions for satire and parody, which our elected officials ought to be able to handle.

At the same time, the better that synthetic media gets, the weaker this policy feels. YouTube has not always strongly enforced its policies around harassment and cyberbullying, particularly in cases where creators are involved in conflicts, and I worry that this policy creates a new battlefield.

One class of creators who will get special protections under the policy are major-label musicians, who will be able to request takedowns of videos that mimic their voices even if used in parodies and satires. (The Verge has a good piece on this today.) We know that YouTube is negotiating with record labels to turn artists’ voices into a generative AI creative tool, presumably with some sort of revenue-sharing deal, and in the meantime this feels like a gift designed to buy some goodwill.

III.

The policy YouTube announced today feels like a first draft, and it surely will evolve before being rolled out to users. Right now, many of the most egregious harms from generative AI are still theoretical. Others are covered by existing policies. The new threat models are still under construction.

Still, it seems noteworthy that the most accessible tools we have for creating synthetic media are more restrictive than the tools we have for distributing it. Meta’s synthetic media policy, which would fit on the back of a napkin, is roughly as permissive as YouTube’s. (TikTok is the only one of the three to ban synthetic videos of non-public figures.)

Perhaps all this will balance out over time. For the moment, though, it seems strange that the likeliest source of harms from AI comes from not from leading AI companies, which will block it at the source of creation, but from open-source tools used to create materials hosted and distributed by tech giants.

If social networks hope to avoid a repeat of the post-2016 tech backlash, they should proceed carefully.

Threads DMs

Threads heads were briefly outraged on Tuesday when Instagram chief Adam Mosseri suggested the company would not build direct messages for the app. In a Threads post, I speculated that DMs probably would arrive eventually, if only because of the adage that all software expands until it includes messaging. Happily, Mosseri replied to me a bit later to say that the company might add DMs eventually, but for now the company is focused on building out features for Instagram DMs.

This is a tricky problem, and one of the more obvious product downsides in launching Threads by leveraging the Instagram network. If Threads winds up feeling like a complement to Instagram, perhaps directing Threads users back there for messages (or mirroring the inbox as a tab inside Threads) is the right approach.

So far, though, Threads feels like its own thing to me. It is heavily text-based, with a strong concentration of journalists and pundits and politicians. The photo- and video-heavy experience of Instagram, which is led by a different category of creators, has yet to see similar traction.

Historically, Meta products that are on track to hit 1 billion users all get their own standalone messaging app: Facebook, Instagram, and (of course) WhatsApp. Efforts to unify those into a single inbox seem to have stalled — probably for good reason.

All of which is to say: if Threads continues on its current trajectory, I suspect at some point Meta will decide building a separate DM experience for the app will be the path of least resistance.

Talk about this edition with us in Discord: This link will get you in for the next week.

Governing

In an important case, a federal judge ruled that lawsuits charging Alphabet, Meta, ByteDance and Snap with harming children can proceed. (Jonathan Stempel and Nate Raymond / Reuters)
Epic v. Google: Testimony from a Google executive revealed the company agreed to pay Samsung $8 billion over four years to make its search engine, voice assistant, and Play Store the default on Samsung’s mobile devices. (Malathi Nayak and Leah Nylen / Bloomberg)
A new filing shows that Donald Trump’s Truth Social and its parent Trump Media and Technology Group have lost at least $31.5 million since launch, taking in $3.7 million in net sales. (Alex Weprin / The Hollywood Reporter)
- The platform is facing “substantial doubt” about its future, the filing says, after suffering negative cash flows and recurring losses from operations. (Drew Harwell / Washington Post)
Google filed a lawsuit against two men, alleging they filed fraudulent takedown notices for hundreds of thousands of URLs, harming the company and its customers. (Andy Maxwell / TorrentFreak)
Almost three dozen venture capital firms signed a set of commitments on how the startups they fund should responsibly develop AI. (Shirin Ghaffary / Bloomberg)
Microsoft and Google will not challenge a law under the European Digital Markets Act that requires them to allow users to switch between competing services more easily. (Foo Yun Chee and Supantha Mukherjee / Reuters)
Adobe will face a challenge by European regulators in its $20 billion Figma acquisition, as regulators reportedly prepare to file anti-competitive charges. (Javier Espinoza / Financial Times)
The EU’s AI Act has created divisions among lawmakers about what practices to ban, how to conduct fundamental rights impact assessments, and potential exemptions for national security practices. (Natasha Lomas / TechCrunch)
X is failing to remove posts that violate its published community guidelines under misinformation and hate speech related to the Israel-Hamas conflict, a report by the Center for Countering Digital Hate found. (Jess Weatherbed / The Verge)
Internet browsing data is being collected with greater detail than previously thought, a new report by the Irish Council for Civil Liberties found. Data on people with sensitive professions, like judges and military personnel, are being used for targeting. (Cristina Criddle / Financial Times)
Nepal is banning TikTok, saying that the platform is “disturbing social harmony and family structures”, citing cases of cyberbullying, financial extortion, and sexual exploitation. (Shan Li and Krishna Pokharel / The Wall Street Journal)

Industry

ByteDance’s revenue surged by more than 40 percent to $29 billion in Q2, closing the gap with Meta through advertising and e-commerce. (Cory Weinberg, Juro Osawa and Jing Yang / The Information)
Shein and TikTok Shop are trying to ditch their “made in China” reputation by courting international sellers and diversifying their supply chain. (Peiyue Wu and Daniela Dib / Rest of World)
Amazon and Snap reached a deal that lets users buy Amazon products from ads on Snapchat, similar to Amazon’s ads on Instagram and Facebook. (Sylvia Varnham O’Regan and Theo Wayt / The Information)
- For Buy with Prime users, Amazon is tacking on package tracking and return services ahead of the holiday shopping season. (Arriana McLymore / Reuters)
WhatsApp chat and media backups stored on Android will soon count towards Google cloud storage limits after a temporary reprieve. (Abner Li / 9to5Google)
Google Deepmind developers say its GraphCast AI model is predicting the weather better than traditional forecasting methods. (Clive Cookson / Financial Times)
Yandex NV is reportedly selling off its Russian business, including its popular search engine, as investors grew wary after the company’s founder criticized the war in Ukraine. (Bloomberg)
A profile of Geoffrey Hinton, the “godfather of AI”, who argues we urgently need to address the ethical implications and potential dangers of the technology. (Joshua Rothman / The New Yorker)
A coder reflects on the challenges and joys of coding, the changing programming landscape, and the potential impact of AI on the future of the profession. (James Somers / The New Yorker)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and deepfakes of your friends: casey@platformer.news and zoe@platformer.news.

How banning one Palestinian slogan roiled Etsy

Casey Newton — Tue, 14 Nov 2023 01:10:44 GMT

Today, let’s talk about a controversial decision inside Etsy to restrict the sale of merchandise with a slogan associated with support for Palestine — and one that many people associate with violent antisemitism. Internal conversations obtained by Platformer highlight how platforms are struggling to develop clear, consistent policies related to the Israel-Palestine conflict — and to manage employees’ feelings as those policies come to light.

On November 2, at an all-company meeting known internally as a “Y’All Hands,” an Etsy employee asked why the platform had restricted sales of merchandise containing the phrase “from the river to the sea.”

It’s a complicated question.

The New York Times dates usage of the phrase to the 1960s, reflecting the desire of Palestinians to reclaim their former land between the Jordan River and the Mediterranean Sea. Today Palestinians and their allies use it both to express “their desire for a right of return to the towns and villages from which their families were expelled in 1948, as well as their hope for an independent Palestinian state, incorporating the West Bank, which abuts the Jordan River, and the Gaza Strip, which hugs the coastline of the Mediterranean,” write the Times’ Karoun Demirjian and Liam Stack.

At the same time, the phrase also appears in the revised 2017 platform of Hamas, the group responsible for the October 7 attacks that killed more than 1,400 Israelis. (The Times notes that the same platform also states that Hamas “could accept a Palestinian state along the borders that were in place before the 1967 war — the same borders considered under the Oslo Accords.”) The Anti-Defamation League, which fights against anti-semitism and discrimination, calls “from the river to the sea” “an antisemitic slogan” that “has the effect of making members of the Jewish and pro-Israel community feel unsafe and ostracized.”

Last week, the US House of Representatives voted 234 to 188 to censure Rep. Rashida Tlaib (D-MI), the only Palestinian-American member of Congress, citing her use of the phrase in their remarks.

As debate continues to rage over the ongoing bloodshed — more than 11,000 Palestinians have been killed since Israel’s military response began, according to Gaza’s Health Ministry — platform policy teams have been confronted with a dilemma. To one group, “from the river to the sea” represents a call for liberty, dignity, and self-determination. To another, it is interpreted as an antisemitic call to annihilate Israel.

In their remarks to employees, Etsy executives highlighted these tensions.

“We know that many people use this phrase with genuine intent and without any violent meaning,” a policy executive told employees in a recording of the meeting viewed by Platformer. “However, it was determined that it violates Etsy’s prohibited items policy for a couple of reasons.”

The first reason was the phrase’s “co-optation by Hamas, which we classify as a dangerous group under our policies,” the executive said. “We also see instances where the phrase has been used alongside celebration or glorification of the Hamas attacks on Israeli civilians.”

Another reason for the decision was more specific to Etsy. On platforms like YouTube and Instagram, where “from the river to the sea” appears in countless videos and creator bios, the nature of the medium gives moderators more context around how the phrase is being used. Watching a video can make it clear whether a user is simply declaring support for the people of Gaza or calling for the end of Israel.

Etsy, on the other hand, is a place for physical goods. The commentary there rarely exceeds the length of a bumper sticker — in fact, Etsy does a brisk business in bumper stickers. Whatever a merchant’s feelings might be about the war, they likely will not be apparent on a slogan-branded coffee mug.

“At Etsy, our priority is keeping our global marketplace safe while embracing a range of perspectives and opinions,” the company told me in a statement. “We recognize that certain topics may be interpreted to mean different things to different groups. In these nuanced cases, we consult industry standards, academic resources, government regulations, and trusted experts across the ideological spectrum to help inform our decisions and take a consistent, objective approach to policy enforcement.”

It also noted that thousands of pro-Palestine products are still available for sale on the platform. A quick search turned up many items bearing the slogan “resistance until reclamation,” for example, and others with “stop the genocide.”

At the same time, the company’s description of its policies elides the fact that merchandise containing the phrase “from the river, to the sea” was permitted on Etsy before the Oct. 7 attacks. It was banned afterward under the company’s prohibited items policy, which includes a prohibition against “items that support, promote, or glorify violence.”

The company told me that the context around the phrase changed after the attacks, and its enforcement changed accordingly. It then notified sellers that their wares now violated the policy. Etsy made no public announcement about the change. Employees only learned about the update after seeing merchants complaining on social media about their products being removed.

In response, some employees complained internally that the policy does not seem to apply in all cases. A policy team decided to leave up a shirt with the Israeli flag bearing the slogan “These colors don’t run, they reload,” for example, generating internal confusion.

Tech companies have not reached a consensus about what place, if any, “from the river to the sea” should have on their platforms. The phrase can be found widely across Facebook, Instagram, YouTube, and TikTok. (TikTok is currently under fire by some members of Congress because the number of videos with the #freepalestine hashtag outnumber those with #standwithisrael.)

I’m told items with the phrase have been banned or heavily restricted on Amazon; the company declined to tell me what its policy is when I asked today. (I couldn’t find any obvious “from the river to the sea” merchandise on the platform, though Fox Business says it did.) eBay also appears to have restricted merchandise with the phrase, though it did not respond to a request for comment.

There is a history of platforms changing their policies around certain words and phrases as the context around them changes. “Save the children,” once among the most innocuous expressions imaginable, took on a new cast when it became a rallying cry for QAnon. (Facebook temporarily blocked the hashtag as a result.)

Just because a phrase has been co-opted, though, doesn’t mean platforms have to ban its use outright. Yoel Roth, former head of trust and safety at Twitter, told me that during his time there the platform worked not to ban phrases and hashtags outright. Instead, it looked for additional context.

“You can get a decent signal in these cases by looking at an account’s longer-term behavior,” Roth told me. “Have they posted potentially dangerous content before? Do they have prior abuse strikes? Are they repeatedly posting borderline/questionable phrases? Do they have multiple accounts doing weird stuff? Do they get reported a lot?”

The idea, he told me, is to “combine content signals with behavioral signals.”

Another idea, one Etsy employee told me, is to make policy decisions in public, and to disclose those decisions when the company does. The company’s top complaint from sellers is that they don’t understand why the company has removed their items from sale, I’m told.

“While we have some legal reasons for not disclosing certain enforcement information,” the employee said, “for many cases, being transparent about our enforcement would do more good than harm.”

In case you missed it, this exchange from Friday’s episode with FTC Chair Lina Khan made the rounds on social media all weekend. Listen to the episode here.

Kevin Roose: All right. Last question — this is something that we ask many podcast guests. What is your P doom, Lina Khan? What is your probability that AI will kill us all?

Lina Khan: Ah, I have to stay an optimist on this one. So I’m going to hedge on the side of lower risk there.

Roose: So are you zero?

Lina Khan: No, no, not zero. Maybe, like, 15 percent.

Casey Newton: And they say there are no techno-optimists in the government.

Governing

Epic v. Google: Documents and testimony showed that Google offered Netflix a special discounted rate of sharing only 10 percent of its in-app payments on Android back in 2017. Netflix declined, though, projecting that it would still lose money. (Sean Hollister / The Verge)
Google v. DOJ: Apple makes 36 percent of the revenue Google earns from search advertising made through the Safari browser, Alphabet’s economic expert testified. The number was supposed to remain confidential. But I’m glad it didn’t! (Leah Nylen / Bloomberg)
- Apple reportedly gets “massive preferential treatment” from Amazon due to a 2018 deal that established an Apple storefront on the e-commerce platform, according to emails revealed as part of the FTC’s Amazon lawsuit. (Chance Miller / 9to5Mac)
Google filed a lawsuit against unnamed individuals in India and Vietnam, alleging that scammers are tricking US small businesses into clicking Facebook ads to download Bard and hitting them with malware that steals social media information. (John D. McKinnon / The Wall Street Journal)
Omegle, the online chat service, shut down. While its founder blamed the stress and expense of operating it, others are pointing out that the service had major issues with child exploitation. (Katie Notopoulos / Insider)
- A lawsuit by a then-underage plaintiff alleging a man in his thirties forced her to take nude photos and videos over a three-year period was settled a week before the shutdown prior. Omegle’s permanent shutdown was required by the settlement, the plaintiff’s attorney said. (Amanda Hoover / WIRED)
A coalition of tech companies, federal agencies, election officials, and researchers aiming to fight foreign propaganda and ensure election integrity has been stymied by repeated attacks from members of the Republican party. (Kevin Collier and Ken Dilanian / NBC News)
The SAG-AFTRA deal that ended the strikes includes protections against the use of AI, including consent and compensation for making digital replicas of actors and disclosure on what the replicas are used for, along with guidelines for synthetic fakes. (Charisma Madarang, Kalia Richardson and Krystie Lee Yandoli / Rolling Stone)
Apple is set to dispute a decision under the European Digital Markets Act that included the App Store under new digital antitrust rules, as well as challenge regulatory scrutiny of iMessage. (Samuel Stolton and Stephanie Bodoni / Bloomberg)
X only has a fraction of the content moderators that Google and TikTok do in the EU, and regulators say they hope X will feel the pressure to catch up. I’m afraid I have some bad news for regulators! (Foo Yun Chee and Supantha Mukherjee / Reuters)
A new tool from Google’s Jigsaw unit, Altitude, aims to help smaller online platforms detect and remove terrorist content by connecting platforms to a shared database of context and information about moderation decisions. (David Gilbert / WIRED)
AI companies including OpenAI and Anthropic have joined the Christchurch Call to Action, a project to suppress terrorist content, started by French president Emmanuel Macron and then-New Zealand prime minister Jacinda Ardern. (Ryan Heath / Axios)
AI chatbots may someday help bad actors build bioweapons, making product guardrails a necessity. (Steph Batalis / Foreign Policy)
Employees at Meta are divided internally over the company’s child safety responsibilities after it lowered the minimum age for Quest users from 13 to 10. (Kalley Huang and Sylvia Varnham O’Regan / The Information)

Industry

Users can now delete Threads profiles without having to delete their Instagram accounts, Instagram head Adam Mosseri said. (Jay Peters / The Verge)
Threads users can now opt out of having their posts displayed on Instagram and Facebook in the “For You on Threads” carousel. (Wes Davis / The Verge)
Tumblr is restructuring its divisions after failing to reach revenue and usage above previous peaks, according to a leaked internal memo, with majority of the 139-person team switching to other divisions. (Adi Robertson / The Verge)
Some TikTok creators in the UK are coordinating a blackout on the app to pressure the company to drop a creator, Anthony Ekpenyong, who was convicted of rape in September, from its creator management program. (Emily Baker-White / Forbes)
Google is reportedly in discussions to invest hundreds of millions in Character.AI, an AI chatbot startup, to help train its models. (Krystal Hu / Reuters)
OpenAI is courting top Google employees with $10 million pay packages and promising more access to computing resources for researchers. (Jon Victor / The Information)
OpenAI announced Data Partnerships, an effort to work with organizations to train AI on public and private data sets, to minimize biases and flaws. (Kyle Wiggers / TechCrunch)
OpenAI CEO Sam Altman says plans to secure more funding from Microsoft to invest in his vision of an artificial general intelligence. (Madhumita Murgia / Financial Times)
Microsoft briefly restricted employee access to ChatGPT, according to an update posted internally, due to security and data concerns. Microsoft later stated the block was a mistake. (Jordan Novet / CNBC)
Meta has reportedly made a preliminary deal with Tencent Holdings to sell a lower-priced version of its VR headset in China, 14 years after Facebook was shut out of the country. (Raffaele Huang, Liza Lin and Salvador Rodriguez / The Wall Street Journal)
WhatsApp is introducing a voice chat feature for large groups that is less disruptive than a group call, letting users join with an in-chat bubble instead of ringing all members of the group. (Aisha Malik / TechCrunch)
Amazon and Meta are partnering up and testing a feature that would let shoppers buy Amazon products directly from ads on Instagram and Facebook. (Spencer Soper and Aisha Counts / Bloomberg)
The new Spatial Video camera feature on the iPhone 15 Pro allows users to record 3D videos that can be played back on the Vision Pro, and it looks amazing, according to this review. (Scott Stein / CNET)
Lidiane Jones is taking over as Bumble’s CEO during a difficult time, as shares of the company have slipped 80 percent since its IPO. But Jones has plans for global expansion, and — of course — AI. (Hannah Murphy / Financial Times)

Those good posts

For more good posts every day, follow Casey’s Instagram stories.

(Link)

Talk to us

Send us tips, comments, questions, and secretly banned phrases: casey@platformer.news and zoe@platformer.news.

Platformer

Why Platformer is leaving Substack

Governing

Industry

Those good posts

Talk to us

Meta seeks to hide harms from teens

Governing

Industry

Those good posts

Talk to us

Substack says it will remove Nazi publications from the platform

What this means for Platformer

Sponsored

Investors are focused on these metrics.

Governing

Industry

Those good posts

Talk to us

Why Substack is at a crossroads

Sponsored

Investors are focused on these metrics.

Governing

Industry

Those good posts

Talk to us

OpenAI’s news blues

Substack and Nazis

Governing

Industry

Those good posts

Talk to us

14 predictions about 2024

Sponsored

Don’t just bank, perform at your highest level.

Governing

Industry

Those good posts

Talk to us

An Epic win jolts Google

Governing

Industry

Those good posts

Talk to us

X restores Alex Jones

Governing

Industry

Those good posts

Talk to us

Meta’s Oversight Board takes on the Israel-Hamas War

Governing

Industry

Those good posts

Talk to us

Google unveils Gemini

Sponsored

Investors are focused on these metrics.

Governing

Industry

Those good posts

Talk to us

How Facebook helps predators find each other

Governing

Industry

Those good posts

Talk to us

Amazon’s Q has ‘severe hallucinations’ and leaks confidential data in public preview, employees warn

Those good posts

Talk to us

Ten strategies for replacing Twitter from people who used to work there

Governing

Industry

Those good posts

Talk to us

The AGs reveal their hand against Meta

Governing

Industry

Those good posts

Talk to us

The OpenAI saga isn’t over just yet