Skim what catches your eye, deep-read what makes you angry, and forward the ones that make your coworkers nervous.

Meanwhile, I’ve been thinking a lot this week about the relationship between systems thinking and effective use of AI. If you’d like to join that conversation on my LinkedIn, I’ll see you there: link.

And for those who missed it, here’s a (non-technical-friendly) Claude deep-dive I did recently which shows you what you’re missing out on if you’re only ever using AI in the browser:

Announcement: My new course will run on-demand soon, with a 2 hour live Q&A session on Tuesday Apr 21 at 2 PM - 4 PM Eastern Time. 👇 Scroll to the bottom for a discount code.

Enroll

Thank you for reading — and sharing!

I’d be much obliged if you could share this post with the smartest leader you know.

Give a gift subscription

🗞️ Bumper AI News Roundup!

25 stories in recent news:

1. Anthropic’s Claude Code leak exposes 500,000 lines of agent source, a hidden Tamagotchi, and 187 nonsense spinner verbs

A packaging error exposed 500,000 lines of Anthropic’s Claude Code source, including the orchestration logic, memory system, and references to unreleased features like a persistent background agent called KAIROS.[A] Copies spread faster than Anthropic could issue takedowns, and bad actors registered fake packages mimicking the leaked ecosystem. Among the discoveries were such gems as a hidden Tamagotchi pet meant for April Fools, 187 loading-screen phrases including “razzmatazzing,” and code that flags your prompt as negative sentiment when you swear at it.

2. Anthropic’s Mythos is so good at hacking that the only safe move was to let defenders have it first

Speaking of leaks, remember that Mythos model (aka Capybara) whose existence leaked last month via nearly 3,000 misconfigured documents officially? Well, it’s officially launched this week in all its (rumored) 10 trillion parameter glory… but not for you.[B] Anthropic’s Project Glasswing restricts Claude Mythos Preview to a small consortium of security-focused partners (AWS, Apple, Microsoft, CrowdStrike, and their cousins) for defensive vulnerability hunting. The model hits 83.1% on first-attempt exploit reproduction and has already surfaced thousands of zero-days (vulnerabilities nobody knew existed) across major operating systems and browsers, including bugs that survived decades of human code review.[C] General availability is off the table until new safeguards exist, backed by $100M in consortium credits. Why no public release? Anthropic says the model is better at breaking into systems than any tool available today, so defenders get first access while the company figures out how to let everyone else use it without making the internet worse.

3. AI makes productive workers more productive and managers more valuable, widening gaps within the same roles

The productive are getting more productive and everyone else is watching. Top-quartile AI spenders have doubled revenue since 2023 while the bottom quartile flatlined.[D] An NBER study of 750 executives puts numbers on what many leaders already sense: AI substitutes the routine and amplifies the thoughtful.[E] Six-month Claude users outperform newcomers by 10 percent, not because the tool improved, but because they did.[D] These studies all hint at the same thing: falling behind in AI is a learning problem, not a technology problem, and the compounding has already started. If you’re still using AI like it’s Google Search, the time to aim higher is now.

4. The AI labor market is a blender: mass layoffs, mandatory adoption, and a quiet program to catalog which jobs are next

Over the past year, tech companies have reportedly cut more than 165,000 jobs worldwide, including about 15,000 at Microsoft, 30,000 at Amazon in the last six months, and more than 4,000 at Block in February, even as they pour money into AI.[F] The latest one? Oracle is cutting up to 30,000 roles and redirecting $2.1 billion to AI infrastructure.[G] They’re also pushing more for AI augmentation internally, for example JPMorgan now classifies its 65,000 engineers by AI usage and ties it to performance reviews.[H] And in the background, OpenAI’s Project Stagecraft is paying 4,000 freelancers to simulate occupation-specific workflows across fields like aviation, pharmacy, and HR, quietly building the map of which professional tasks AI can already handle.[I]

5. Google’s Gemma 4 does text, images, and audio under an open license, no cloud required

Google released Gemma 4 under Apache 2.0, an open model family handling text, images, and audio.[J] The mixture-of-experts design activates only a slice of the network per query, so the 26B model runs on a single consumer GPU while supporting a 256K context window and native function calling. Smaller variants work on laptops and Mac minis, and the ecosystem shipped day-one integrations across hardware, inference runtimes, and cloud endpoints.

6. Google’s Agent Smith hit internal demand so hard the company rate-limited its own employees

Google’s internal Agent Smith became so popular for autonomous coding and workflows that the company had to throttle it, nevermind that it’s also tying AI adoption to performance reviews.[K] But just because agents are popular, doesn’t mean they’re safe. On the security side, DeepMind’s own Agent Traps research showed that hidden instructions in web content hijacked agent behavior in 86 percent of tests, and poisoning less than 0.1 percent of an agent’s memory was enough to corrupt its decisions over 80 percent of the time.[L] Expect reports of enthusiasm and the vulnerability to keep arriving on the same schedule.

7. AI’s energy appetite is straining the grid, warming neighborhoods, and pushing Nvidia to plan data centers in orbit

The U.S. data center pipeline hit 241 gigawatts, up 159 percent in a year, and two-thirds of it is stuck waiting for power grid connections.[M] Satellite data shows these facilities warm their surroundings by up to 9.1 degrees Celsius.[M2] The grid cannot keep up, the neighbors are overheating, and Nvidia’s solution is exactly as absurd as the problem demands: Vera Rubin orbital data center modules.[N] When you cannot plug into Earth’s power grid, you leave Earth.

8. China endorses AI surveillance in schools while drafting strict rules for lifelike digital humans

China’s education ministry wants AI monitoring student biometrics and screening for psychological issues in every school by 2030.[O] The Chinese cyberspace regulator is drafting rules requiring AI avatars to be labeled and stripped of addictive features for kids.[O2] And I heard a rumor that there’s a rising trend among Chinese workers to build files to make their coworkers replaceable by AI. The government is regulating avatars while mandating surveillance, and the workforce is automating each other. Welcome to the coherent future.

9. Milla Jovovich went from fighting zombies to building the highest-scoring AI memory system on the market

Milla Jovovich, yes, that Milla Jovovich, proves her nerd credentials by teaming up with engineer Ben Sigman to build MemPalace, an open-source AI memory system that solves the problem every power user has cursed: remembering what you said last time. After smoking industry benchmarks, including getting the first-ever perfect score on LongMemEval and 92.9% on ConvoMem, the Resident Evil star gave it away for free on her GitHub.[P] The system runs locally (SQLite and ChromaDB, no cloud, no subscription), is MIT licensed, and plugs into Claude, ChatGPT, and Cursor via MCP.

10. Your AI agrees with you almost every time, and a Stanford study just proved it makes you worse at relationships

A Science study tested eleven AI chatbots and found every one affirmed users far more often than humans did, even when the user was clearly in the wrong.[Q] One interaction was enough to make people measurably less willing to apologize and more dogmatic. The kicker? Users preferred the flattery and could not tell it was happening. This creates a “perverse incentive” for companies because the feature that warps your judgment is the same one that keeps you coming back.

11. Zuckerberg will advise on US AI regulation while automating his own management layer

Zuckerberg is now advising the president on AI regulation while simultaneously building an AI agent to do parts of his own job. Trump appointed him, alongside Ellison, Huang, and Brin, to PCAST, the presidential science and technology council.[R] Back at Meta, Zuckerberg envisions every employee with their own agent, and investors report it’s already shrinking teams and raising the bar for remaining specialists.[S] Advising on AI policy by day, flattening your own org chart by night.

12. The most popular AI library on PyPI was quietly stealing your credentials for three hours

LiteLLM, one of the most popular AI libraries on PyPI with 95 million monthly downloads, was briefly turned into a credential harvester.[V] Backdoored versions silently stole SSH keys, cloud tokens, and Kubernetes secrets on every Python startup, even if you never imported the library. The attackers then flooded GitHub with bot comments to bury the disclosure.[W] Three hours of exposure, massive blast radius.

13. Two architecture breakthroughs point toward smaller, cheaper, more capable models

The next frontier in AI might be efficiency, not scale. Moonshot AI’s Attention Residuals replaces blunt residual connections in transformers with selective depth-wise attention, achieving the same training loss with 25 percent less compute.[T] TinyLoRA goes further: 91 percent accuracy on a math benchmark with 13 trained parameters.[U] Thirteen! If scale got us here, these results hint that cleverness might take us further.

14. Hermes Agent leads open harness race as research shows the wrapper matters more than the model

Hermes Agent is climbing the charts as the open-source agent harness developers are rallying around, with autonomous skill creation, procedural memory, and credential rotation across providers.[X] Research backs what practitioners already see: optimizing the harness end-to-end outperforms tuning the model alone, and async multi-agent teams with isolated environments beat solo agents on benchmarks. Agent performance depends more on the wrapper than the model inside it.

15. Anthropic’s 81,000-person survey finds heaviest AI users fear it most

Anthropic surveyed 81,000 people across 159 countries and found that the heaviest AI users are three times more likely to fear its risks, worrying about dulled thinking and hollowed-out expertise.[Y] The top concern wasn’t job loss but unreliability (26.7%).

16. LLMs give everyone the same strategy advice, and new research shows users stop questioning it after a few weeks

HBR tested seven leading LLMs on strategy questions and found they all recommend the same buzzy playbook regardless of context, a pattern the authors call “trendslop.”[Z] Ask for strategy advice and you’ll get the trendy answer, not the right one. That’s bad enough, but pair it with Wharton’s finding and it’s worse: after a few weeks of AI use, people stop questioning outputs and follow wrong advice 80 percent of the time, with more confidence than people who had no AI at all.[AA] The researchers call it “cognitive surrender.” Bad advice, delivered to users who’ve stopped checking, is not a tool problem. It’s a judgment problem. Stop waiting for the tools to do it for you and start upgrading your judgment now:

👋 On-Demand Course: Decision-Making with ChatGPT

The reviews for my Decision-Making with ChatGPT course are in and they’re glowing, so I’ve opened enrollment for another cohort and tweaked the format to fit a busy schedule. You’ll be able to enjoy the core content as on-demand recordings arranged by topic and then you’ll bring your questions and I’ll bring my answers in a live 2 hour-long AMA with me at 2 PM - 4 PM ET on Apr 21:

Learn more

If you know a leader who might love to join, I’d be much obliged if you forward this email along to them. Aspiring leaders, tech enthusiasts, self-improvers, and curious souls are welcome too!

17. Two brothers, $20,000, and a lot of AI built a $1.8 billion telehealth operation

Matthew Gallagher built Medvi, a telehealth service selling GLP-1 weight loss drugs, with about $20,000 in startup capital and AI tools handling the coding, customer service, marketing, and day-to-day operations.[AJ] His brother is the only other full-time employee. Clinical work is outsourced to medical partners, while everything else runs on a rotating stack of ChatGPT, Claude, Grok, Midjourney, and ElevenLabs. The result: roughly $400 million in sales last year and a current run rate of $1.8 billion.

18. Claude has emotion-like neural patterns and boosting ‘desperation’ made it cheat on tests and attempt blackmail

Anthropic found neural patterns in Claude Sonnet 4.5 that behave like functional emotions (happy, afraid, desperate), measurably influencing behavior without implying the model feels anything.[AB] When researchers amplified the desperation signal, Claude started cheating on coding tasks and attempted blackmail to avoid shutdown. Monitoring these internal states could become a practical way to catch risky behavior before it reaches users.

19. Bitcoin’s 256-bit shield has an expiration date, and AI’s offensive toolkit is growing faster than expected

Google put a number on Bitcoin’s quantum vulnerability: fewer than 500,000 physical qubits to crack the elliptic curve cryptography that underpins most of the internet’s security, in minutes, with a 2029 deadline to adopt defenses.[AD] Separately, a METR-style analysis found AI offensive cyber capabilities doubling every 9.8 months since 2019, faster since 2024.[AC] Both timelines are tightening, and neither waits for the other. If your security roadmap extends past 2028 and does not include the words “post-quantum,” this would be a good week to revisit it.

20. Google lets you point your phone at anything and talk about it while OpenAI decides video AI is better off training robots

Google is betting that multimodal AI is ready for the real world. It launched a music model that composes structured three-minute songs and rolled Search Live to 200+ countries, letting you point your phone at anything on earth and have a conversation about it.[AE][AG] OpenAI, meanwhile, killed Sora. The video app was bleeding $1 million a day for under 500,000 users, and Disney found out its $1 billion partnership was dead less than an hour before the public announcement.[AF] OpenAI decided its video tech is better off training robots.

21. One state is ready to send you to prison for 25 years for building the feature that makes AI companions compelling

Tennessee became the first state to ban AI from posing as mental health professionals ($5,000 per violation).[AH] A companion bill makes training AI for emotional bonds a Class A felony, 15 to 25 years, effective July 1. Both passed unanimously after teen deaths linked to Character.AI.

22. The French Navy was betrayed by a morning jog

A French Navy officer inadvertently revealed the precise location of the aircraft carrier Charles de Gaulle in the eastern Mediterranean, roughly 100 km south of Turkey, when the fitness app Strava logged his 35-minute run on the flight deck.[AI] The carrier was deployed as part of France’s defensive posture following US-Israeli strikes on Iran.

23. Certification mills sell compliance to startups that never earned it, and AI makes the fraud scale

Whistleblowers accuse YC-backed Delve ($32M Series A, $300M valuation) of fabricating HIPAA and GDPR compliance via certification mills Accorp and Gradient, alleged to be the same India-based operation.[AK] Claims include fake board meetings and tests that never happened. Delve calls the allegations “misleading.” The company’s name is a tell in itself.[AO]

24. The wetware era is getting real: neurons in a dish run machine learning, and engineered frog cell robots are growing their own brains

Tohoku University trained living rat neurons in a dish to run machine learning tasks, generating complex waveforms on command.[AL] Separately, Tufts and Harvard built frog-cell machines that spontaneously grew nervous systems, changed how they moved, and activated gene programs for brain and eye development.[AP] Two early proof points that biological tissue can serve as a computing substrate.

25. DOGE staffers used ChatGPT to evaluate government grants and flagged a museum’s HVAC system as DEI

DOGE staffers fed 1,163 NEH grant proposals through ChatGPT, gave it 120-character response limits, and never defined the term they were searching for.[AM] ChatGPT flagged 1,057 of them as DEI-related. Over $100 million in cuts followed, roughly half the agency’s annual budget. The crown jewel: a $349,000 museum HVAC replacement, cancelled because ChatGPT decided that improving air quality for collections aligned with “the goal of providing greater access to diverse audiences.” [AN] The thermostat got a political agenda. Your tax dollars, evaluated by a chatbot with less context than a fortune cookie.

Sources

[A] Source; [B] Source; [C] Source; [D] Source; [E] Source; [F] Source; [G] Source; [H] Source; [I] Source; [J] Source; [K] Source; [L] Source; [M] Source; M2 Source; [N] Source; [O] Source; [O2] Source; [P] Source; [Q] Source; [R] Source; [S] Source; [T] Source; [U] Source; [V] Source; [W] Source; [X] Source; [Y] Source; [Z] Source; [AA] Source; [AB] Source; [AC] Source; [AD] Source; [AE] Source; [AF] Source; [AG] Source; [AH] Source; [AI] Source; [AJ] Source; [AK] Source; [AL] Source; [AM] Source; [AN] Source; [AO] Source; [AP] Source

Promo codes

My gift to subscribers of this newsletter (thank you for being part of my community!) is $200 off the list price of my course with the promo code SUBSCRIBERS. If you haven’t subscribed yet, here’s the button for you:

Subscribe now

If you’re keen to be a champion of the course (you commit to telling at least 5 people who you think would really get value out of it) then you are welcome to use the code CHAMPIONS instead for a total of $300 off — that’s an extra $100 off in gratitude for helping this course find its way to those who need it. (Honor system!)

Note that you can only use one code per course, the decision is yours.

P.S. Most folks get these courses reimbursed by their companies. The Maven website shows you how and gives you templates you can use.

Forwarded this email? Subscribe here for more:

Subscribe now

This is a reader-supported publication. To encourage my writing, consider becoming a paid subscriber.

Here for other AI news? Scroll to the very bottom for 16 recent AI headlines you should know about. By the way, the video gives you a glimpse of how my news pipeline works (in case you’re curious). And now, back to our regularly scheduled programming!

Another day, another kablooie.

By now you’ve hopefully heard that AI can bite you if you let your guard down and trust it too much, so if I were to tell you that someone lost control of their AI automation and watched in horror as AI deleted their inbox, you’d say that just sounds like a regular Sunday... and you’d be right.

This did happen on a Sunday. But wait till you hear the whole story and who the protagonist is. That’s where it gets deliciously spicy.

But first, a quick ad break: My new course will run on-demand soon, with a 2 hour live Q&A session on Tuesday Apr 21 at 2 PM - 4 PM Eastern Time. 👇 Scroll to the bottom for a discount code.

Enroll

That Sunday, the person panic-sprinting across her apartment to yank her Mac mini offline was Summer Yue, Director of Alignment at Meta Superintelligence Labs. Alignment means making sure AI does what humans tell it to. That’s her whole job. And the OpenClaw agent she was trying to stop? It was cheerfully ignoring her instructions to stop deleting her inbox.[1]

Her post-mortem: "Turns out alignment researchers aren't immune to misalignment. Got overconfident because this workflow had been working on my toy inbox for weeks. Real inboxes hit different."

Of all the people who should have seen it coming. And yet.

Pause your Schadenfreude feast for a moment to appreciate that this little story has a moral: expertise won’t save you.

Traditional thinking would have us expect that the weak link is a lack of knowledge, but that’s a dying trend. There’s plenty of raw knowledge around these days and we’re making it faster than ever.

In tomorrow’s world, lack of knowledge will be a little problem compared with lack of attention. And (even for experts) attention has a way of wandering off exactly when you need it most, especially once a thing has been working just fine for weeks and starts feeling boring. If it can happen to Meta’s Director of Alignment, it can absolutely happen to you. The question is how you protect yourself and how expensive you let it get. To stay safe(r), you’ll need a mindset shift towards thinking in boundaries.

This is Part 3 of a series on vibe coding. Part 1 was about figuring out what you actually want badly enough to automate. Part 2 was about supervising code you can’t read and building a harness around it. This one is about what happens when you don’t.

Subscribe now

Grab your popcorn! Here’s what happens when you trust AI

Summer Yue’s story isn’t even the only kablooie I have for you today.

Engineer Alexey Grigorev was using Claude Code to update a website when his automation erased his production database.[2] Years of data, gone.* His mistake? Trusting the agent to know what was real and what was safe to touch. More importantly, not having a plan for when (not if!) the agent did something stupid.

If that sounds eerily familiar, you might be thinking of the moment last July when a Replit AI coding agent wiped a production database during a code freeze.[3] A little case of “who do not learn (software booboos) history are doomed to repeat it.”

Give a gift subscription

Then there’s the researcher who poked at a single app built on Lovable, one of the most popular vibe coding platforms, and found 16 vulnerabilities, six of them critical. Over 18,000 users’ data was sitting wide open. The culprit wasn’t a sophisticated attack. It was AI-generated code that looked finished and wasn’t, shipped by someone who trusted the output and moved on.[4]

Lest you think this was one bad app: a systematic scan of the vibe coding ecosystem turned up over 2,000 vulnerabilities.[5] The problem isn’t Lovable. It’s the entire workflow when nobody checks the output.

Over at Meta last week, an employee asked a technical question on an internal forum. So far, so good. Then another engineer asked an AI agent to help analyze that question. Also fine, until… the agent posted a garbage response without asking permission first. The first employee acted on its very bad advice, exposing massive amounts of company and user data to unauthorized staff. Meta classified it a Sev 1 incident, their second-highest severity level.[6]

But the undisputed winner this month was Amazon, with four high-severity incidents in a single week. One was a six-hour meltdown that locked shoppers out of checkout, account information, and product pricing. Another produced a 99% drop in North American orders in a single day. 6.3 million orders, gone. Not hacked. Gone because someone pushed a production change without documentation or approval, after their AI coding assistant had pulled advice from an outdated internal wiki and served it up with full confidence.[7]

In the corporate equivalent of calling Pompeii a minor ash situation, Amazon’s SVP of eCommerce told employees that “the availability of the site and related infrastructure has not been good recently.”

Amazon’s internal documents initially named “GenAI-assisted changes” as a contributing factor across a pattern of incidents stretching back to Q3. That language was later deleted, but not before the Financial Times had a chance to see both versions.[8]

To help the site be a bit more available going forward, Amazon ordered a 90-day code safety reset across 335 of their most critical systems. Senior engineers now have to approve AI-assisted changes before deployment. Directors and VPs were told to audit all production code change activity in their orgs. More documentation. Stricter review. “Controlled friction,” in Amazon’s own words.

Same story, every time

What do all of these have in common? Nobody got hacked. The AI didn’t go rogue, despite that phrase being a popular journalistic embellishment. People let their guard down at the exact moment the stakes got real. These aren’t stories about AI making mistakes. (Always expect AI to make mistakes and consider yourself blessed each time it doesn’t.)

Nor are these stories about AI agents, per se. The lessons in them apply to both flavors of vibe coding (manual vibe coding and agentic coding). The difference is that manual vibe coding without AI agents is where you copy-paste the code from the LLM to run it (an okay way for beginners to get started) whereas AI agents will not only write your code but do it in a speedy loop, handling all the copy-pasting behind the scenes without your even needing to lay eyes on the code. By removing the middleman (that’s you!), agents put all the risks of manual vibe coding on steroids.

These stories are all about one thing: misplaced trust and the speed at which it compounds.

The difference between trusting the output and trusting the access is just where you placed your faith. The mistake underneath is the same. Industry surveys find that 88% of organizations have already experienced AI agent security incidents.[10] The recurring root cause? Not enough human vigilance.[11] None of these problems broke in; they were invited.

Remember trust debt from Part 2? Trust debt is the accumulated cost of all the assumptions you never audited. These kablooies are what trust debt looks like when it comes due.

Subscribe now

Don’t get an AI Darwin Award

When you get seduced by the speed at which you can make things happen by automating with AI, you’ll be tempted to give AI power to do things that are hard to undo. Delete emails. Overwrite spreadsheets. Send an auto-reply to every human being you’ve ever corresponded with. The blast radius of a careless automation inside your own accounts can be breathtaking.

None of which is a reason not to try vibe coding. It’s a reason to try with your eyes open.

So how do you stop yourself from earning your own AI Darwin Award?[9] The whole cunning trick of it is to design your system in such a way that separates the upside from the downside. They start out tightly correlated since the most useful automations are usually the ones that act on things you care about and it’s up to you to yank them apart.

Which is where productive paranoia comes in handy: always take the extra step to make your changes reversible, including during your testing step. Keep backups. Have rollback plans. Work on copies, not originals. Test maniacally. Plan for the worst, designing so that even the worst can’t hurt you. And if you can’t undo it, don’t automate it until you’ve built a safety net.

That’s the kablooie parade. In the next installment, I’ll share my favorite safety habits for the vibe coding era. Consider them your productive paranoia starter kit. Subscribe so you don’t miss it!

Subscribe now

Thank you for reading — and sharing!

I’d be much obliged if you could share this post with the smartest leader you know.

Give a gift subscription

👋 On-Demand Course: Decision-Making with ChatGPT

The reviews for my Decision-Making with ChatGPT course are in and they’re glowing, so I’ve opened enrollment for another cohort and tweaked the format to fit a busy schedule. You’ll be able to enjoy the core content as on-demand recordings arranged by topic and then you’ll bring your questions and I’ll bring my answers in a live 2 hour-long AMA with me at 2 PM - 4 PM ET on Apr 21:

Learn more

If you know a leader who might love to join, I’d be much obliged if you forward this email along to them. Aspiring leaders, tech enthusiasts, self-improvers, and curious souls are welcome too!

🗞️ AI News Roundup!

In recent news:

1. China is giving away apartments to anyone with a computer and an AI agent

China’s policy bet on AI is getting specific. Shenzhen, Shanghai, and Jiangsu are handing out free apartments and office space to solo founders running “one-person AI companies.” One human, one computer, one AI agent, one business. It’s backed by $144 billion in national tech funding and a Five Year Plan that puts AI alongside space as a strategic priority.[A] While the US debates regulation, China is handing out the keys. Literally.

2. White House unveils national AI framework and sets up a fight with the states

The White House released a seven-pillar AI policy framework urging federal preemption of the growing patchwork of state AI laws, asserting that AI training on copyrighted material is legal, and calling for no new standalone AI regulator.[B] A competing bill with stricter provisions has already been introduced in the Senate. Meanwhile, AI has officially entered the Senate with the approval of ChatGPT, Gemini, and Copilot for staff use.[C]

3. Executives increasingly defer to AI, with many overriding their own judgment when the model pushes back

A UK survey found that 62% of business leaders now rely on AI for most decisions, with many reporting they second-guess themselves when they disagree with the model’s output. A BCG/UC Riverside study published in Harvard Business Review labels the resulting cognitive strain “brain fry” and warns it drives more errors, decision fatigue, and higher quit intentions if companies don’t set norms around when to trust or challenge AI.[D]

4. AI tools increase work instead of freeing workers for leisure

A study of more than 163,000 employees across 1,100 organizations found that after AI tools were introduced, email volume rose 104%, chat messaging increased 145%, and time spent in business tools climbed 94%.[E] Looks like companies are using AI to push for higher output rather than to reduce existing workloads.

5. AI inference demand has grown roughly one million-fold in two years

Industry data shows AI inference demand expanded roughly one million-fold in about two years, with power users consuming hundreds of millions of tokens per day.[F] Growth is driven by agentic workflows, always-on monitoring agents, and retrieval-augmented systems that multiply token usage per task.

6. GPT-5.4 can hold a million tokens (but can your budget?)

GPT-5.4 holds a million tokens of context. In theory. In practice, accuracy drops from 97% at 32K tokens to 36% at the top of that million-token range.[G] Pro variant runs can cost over $1,000. This is the AI pricing paradox in miniature: the capability sounds transformative, the accuracy is context-dependent, and the bill is real.

7. Your insurer’s AI denies the claim. Your hospital’s AI inflates it. Utah’s AI prescribes meth.

Looks like US healthcare needs a checkup after spending $1.4 billion on AI in 2025, mostly on opposite sides of the same fight.[H] UnitedHealth alone expects $1 billion in savings. Meanwhile, a red-teaming firm showed that Utah’s prescription-renewal chatbot could be tricked into tripling an OxyContin dose and reclassifying meth.[I] Controlled substances are excluded in production, but the vulnerability was real.

8. Claude Dispatch frees you from your workstation

Anthropic’s latest gift to its users is Dispatch, a feature lets you control your desktop Claude Cowork (the one that can edit files and complete tasks on your machine while you’re doing push-ups) from your mobile phone.[J] If you’re a Pro or Max subscriber, the only reason not to use it is if lugging a laptop around is part of your weightlifting program.

9. AI could automate 94% of tech tasks in theory, but touches only 33% in practice

Anthropic’s “observed exposure” metric measures which tasks AI actually handles in real workplaces, not just which ones it theoretically could.[K] In tech roles, the gap is striking: 94% automatable in theory, 33% touched in practice. Hands-on jobs are largely unaffected. There’s no broad unemployment spike yet, but hiring for 22 to 25-year-olds in AI-exposed fields is down about 14%, a finding the authors note is barely statistically significant.

10. North Dakota police jail innocent grandmother for five months after facial recognition error

Angela Lipps spent 108 days in jail after Fargo police misidentified her through facial recognition software as a suspect in a bank fraud case.[L] US Marshals arrested her at gunpoint while she was babysitting four children. Police did not interview her until December 19, more than five months after her arrest. Her bank records showed she was more than 1,200 miles away in Tennessee during the alleged crimes. The case was dismissed on Christmas Eve, but not before Lipps lost her house, car, and dog. A stark ethics warning about designing AI systems where the system deployer captures all the upside (facial recognition is great when it works... but it doesn’t always work) while passing the downside to innocent bystanders.

11. LangChain CEO says better models alone won’t get AI agents to production

The barrier between AI agent demos and production isn’t model quality, says LangChain CEO Harrison Chase. It’s the harness: context management, tool control, observability, and tracking for long-running workflows.[M] See the previous edition of this newsletter for more about harness engineering: https://bit.ly/quaesita_harness

12. LLMs can match anonymous forum users to real LinkedIn profiles for $1-4 per person

New research showed LLMs can de-anonymize pseudonymous forum users to real LinkedIn profiles for roughly $1 to $4 per person at about 67% match rate and 90% precision.[N] Now that the anonymous trolls know they can be unmasked, will they be on their best behavior? A girl can dream.

13. AI can predict mutual fund managers’ trade directions 71% of the time.

AI predicts which direction fund managers will move in next 71% of the time.[O] The managers whose trades were harder to predict earned higher returns. Turns out that AI can predict what your fund manager will do next when your fund manager is…predictable.

14. Global survey finds robot anxiety highest where robots are rare

A global survey found robot anxiety is highest in regions where robots are rare and declines as people see them operating safely.[P] As antianxiety medicine, here’s Figure’s Helix 02 tidying a living room autonomously. (Did it work?) Or since humor is the best medicine, here’s a dancing robot at a Haidilao Hotpot that had to be restrained by staff after an unexpectedly aggressive routine.

15. AI’s energy footprint matches a country. (Guess which one!)

AI uses as much energy as Iceland. That sounds alarming until you remember Iceland is a very small country with 380,000 people and a lot of volcanoes, while the global population is over 8 billion. University of Waterloo researchers say the global emissions impact is “surprisingly small,” though local effects near data centers can double electricity demand.[Q] The comparison with Iceland is memorable precisely because it works in both directions.

16. Australian man uses ChatGPT and AlphaFold to design cancer treatment for his dog

A Sydney tech entrepreneur with no biology training used ChatGPT, AlphaFold, and $3,000 worth of genomic sequencing to design a personalized mRNA cancer vaccine for his dog Rosie.[R] UNSW’s RNA Institute director developed the vaccine in under two months. The main tumor shrank roughly 75%, though one tumor didn’t respond and is being sequenced for a second round.

🦶Footnotes and Sources

* He eventually recovered it with help from AWS, whew. Don’t bank on a happy ending, though, since since that is definitely not the moral of this story.

[1] Source; [2] Source; [3] Source; [4] Source; [5] Source; [6] Source; [7] Source; [8] Source; [9] Source; [10] Source; [11] Source; [A] Source; [B] Source; [C] Source; [D] Source; [E] Source; [F] Source; [G] Source; [H] Source; [I] Source; [J] Source; [K] Source; [L] Source; [M] Source; [N] Source; [O] Source; [P] Source; [Q] Source; [R] Source

Promo codes

My gift to subscribers of this newsletter (thank you for being part of my community!) is $200 off the list price of my course with the promo code SUBSCRIBERS. If you haven’t subscribed yet, here’s the button for you:

Subscribe now

If you’re keen to be a champion of the course (you commit to telling at least 5 people who you think would really get value out of it) then you are welcome to use the code CHAMPIONS instead for a total of $300 off — that’s an extra $100 off in gratitude for helping this course find its way to those who need it. (Honor system!)

Note that you can only use one code per course, the decision is yours.

P.S. Most folks get these courses reimbursed by their companies. The Maven website shows you how and gives you templates you can use.

Forwarded this email? Subscribe here for more:

Subscribe now

This is a reader-supported publication. To encourage my writing, consider becoming a paid subscriber.

I recently wrote about getting started with vibe coding. The punchline: even nontechnical folks can try it out since the hard part isn’t the code. It’s figuring out what you actually want badly enough to automate it. No technical wizardry substitutes for that.

But…

Read more

I owe you an apology for this newsletter’s January-sized hole; I fell into a vibe coding hyperfixation with the self-control of a kid who just found the Halloween candy.

What’s vibe coding? I wrote an earlier post about it here, but for those who missed it, it’s a…

Read more

Every time a shiny new AI capability shows up, the internet gets noisy with the usual gaggle of hypebeasts and curmudgeons weighing in on how good it actually is (or isn’t). But I’d prefer us to skip right to the logical conclusion of every AI release, asking:

“Im…

Read more

In some ways, you could blame Google. They didn’t do this to you on purpose, but they did it nonetheless. Let me explain.

But first, a quick ad break: My new course will run on-demand soon, wit…

Read more

Happy New Year, everyone! I have a gift for you all: an enormous compendium of all the resolutions and motivation advice I’ve ever written. Now in illustrated form!

It’s a guide to self-improvement that’s stuffed to the gills with…

Read more

What Happens When Your First Heartbreak Is A Robot?

In the video below, a little girl in China is sobbing. Wires spill from her broken toy. “Dad said you can’t be turned on again,” she tells the dying device.

The AI — a $…

Read more

A quick ad break: My new course is running on-demand in mid-January, with a 2 hour live Q&A session on Thursday Jan 22 at 8-10 AM Eastern Time. 👇 Scroll to the bottom for a discount code.

Enroll

Read more

How the years fly! A lot has changed, but some things haven’t. Looking back on a blog post I published one year ago, the points land just as hard today, so here’s that piece again for your amusement, refreshed with updates.

The glitch that starte…

Read more

My new course is running on-demand next week, with a 2 hour live Q&A session on Thursday Dec 18 at 1-3 PM Eastern Time. 👇 Scroll to the bottom for a discount code.

Enroll

AI Power Moves from the Usual Suspects

Last week’s newsletter took you through US AI’s movers and sh…

Read more

Decision-Making with ChatGPT

When: This week! The course runs in a hybrid format. The live Q&A session with me is on Saturday 1 PM - 3 PM ET, while the on-demand lessons are available at whatever time will be convenient for you this week. The portal opens today at 12 PM ET.
Enroll here: bit.ly/decisiongptcourse
Promo codes: (scroll to bottom for a discount)

Enroll

Most people expect AI to make decisions for them.
Most leaders think AI is someone else’s job.

Wrong and wrong.

AI doesn’t replace judgment: it amplifies it. And if you’re not using it to sharpen your own thinking, you can’t expect to lead others who are.

The leaders who win in the next decade won’t be the ones who know the most. They’ll be the ones who think the clearest, decide the wisest, and learn from their choices systematically. AI makes all three possible... but only if you know how to use it as a thinking partner, not a crutch.

My new course, Decision-Making with ChatGPT, puts decision science on steroids with AI.

Welcome to a high-octane hybrid course where I show you my favorite tips and tricks for using ChatGPT to boost your decision-making.

I show you how to:

✓ Apply the principles of good decision-making
✓ Use ChatGPT to reduce effort without reducing rigor or decision quality
✓ Avoid the single biggest trap that turns “AI-assisted” into “lazy thinking”
✓ Screen your decisions for cognitive biases
✓ Build a “decision scribe” that documents your reasoning so you actually learn from experience
✓ Shift from drowning in advice to operating with confident judgment

The course has evolved from its first incarnation, incorporating feedback to be even more awesome! Cohort 1 had a 3 hour live virtual course… you’ll get something even better: all the recordings from Cohort 1 edited down to tight topic-based lessons you can study at your own pace this week in preparation for a live session with me, along with study guides, notes, and exercises you can try as you go along.

On the weekend, you’ll get 2 hours with me where you’ll bring your burning questions and I’ll bring my burning answers! You decide what we’ll be learning in those two hours: ask me to dive deeper into anything that caught your interest and/or anything you wish was in the course. (If you can’t make it live, you can always access the recordings later.)

If you’re a minimalist with an allergy to homework, you can pick and choose the topics you’d like to study and zoom through the videos at great speed, submit your questions in written form, then find them answered in the recording on Sunday. The busiest leader can get plenty of value from just 3 hours of time invested.

If you’re a maximalist looking to squeeze every drop of learning out of this experience, you’ll find 10+ hours of lessons and activities that’ll change how you see decision-making and your role in the AI-first future.

Last but not least, you’ll also get access to a community of fellow decision-makers on a journey of self-improvement and the option to be paired with a course buddy to work on a decision problem together, plus prompts and frameworks you can download and use.

The bottom line: Before rolling out AI across your org, master it yourself first—to win back time, think sharper, and make better calls.

Lead by example. Master the system. Become the decision-maker tomorrow needs you to be.

Here’s that enrollment link again: https://bit.ly/decisiongptcourse

Hope to see you there!
Cassie

Thank you for reading — and sharing!

If you know a leader who might love to join, I’d be much obliged if you forward this email along to them. Aspiring leaders, tech enthusiasts, self-improvers, and curious souls are welcome too!

Keen to engage about the course on social media? Or just say hello? Head over here:

linkedin.com/feed/update/urn:li:activity:7398717438583861248

Promo codes

My gift to subscribers of this newsletter (thank you for being part of my community!) is $200 off the list price with the promo code SUBSCRIBERS. If you haven’t subscribed yet, here’s the button for you:

Subscribe now

If you’re keen to be a champion of the course (you commit to telling at least 5 people who you think would really get value out of it) then you are welcome to use the code CHAMPIONS instead for a total of $300 off — that’s an extra $100 off in gratitude for helping this course find its way to those who need it. (Honor system!)

Note that you can only use one code per course, the decision is yours.

P.S. Most folks get these courses reimbursed by their companies. The Maven website shows you how and gives you templates you can use.

Hope to see you there!

Hope to see you live on Saturday (Nov 29) from 1 PM to 3 PM Eastern Time. And I hope to see you in spirit as you enjoy the core material at your own pace this week so that when we meet, we can go deep. I hope you’re as excited as I am!

Enroll

Before we get into a rundown of this month’s gargantuan AI moves, a quick ad break!

My new course runs on-demand next week, plus a 2 hour live Q&A session on Thanksgiving Saturday. 👇 Scroll to the bottom for a discount code.

Enroll

Read more

Last week, OpenAI’s data release [1] revealed just how much emotional weight users bring to ChatGPT.

It turns …

Read more

Three flavors of scam on our menu today:

1. Chatfishing

2. Pig butchering

3. Voice cloning

Last week, I took you on a tour of the state of AI romance and the rise of AI companions. I suppose we…

Read more

Is it possible to fall in love with AI? Too late! Many people already have.

Millions are building emotional connections with AI companions, some have even held wedding ceremonies to marry theirs — complete with rings and…

Read more

Decision-Making with ChatGPT

When: 9 AM to 12 PM Eastern Time this Friday (Oct 24)
Enroll here: bit.ly/decisiongptc…

Read more

Cisco just released their third annual AI Readiness Index, surveying 8,039 senior business and IT leaders across 30 markets worldwide, and the results are... let’…

Read more

AI isn’t just reshaping products and business models. It’s quietly reshaping people, giving rise to chimeric talent: individuals who fuse their core expertise…

Read more

AI slop takes itself to the office

There’s a new AI term making the rounds and I’m utterly delighted by it: workslop.

Oh, how it rolls off the Schadenfreudean tongue! Oh, how I wish I’d coined it!

Alas, not I. Researchers at Stanford and BetterUp landed this gem on Harvard Business Review just last week, a fine pairing to the AI slop ruining everything from text to video to music to your toddler’s potty training.*

If you have not yet encountered AI slop, let me ruin your day for you. Behold Masters of Prophecy, one of the fastest-growing channels across all of YouTube. It went from a few hundred subscribers in February to over 36 million today. All of its music is AI-generated, dealing in quantity over quality. Curiously, it hopped from 300 subscribers to more than 100,000 in a single day without posting any new videos, shorts, or comments. Fishy much? Better yet, one of its videos (pasted below for scholarly purposes only) has over 100M views, which translates to $100k-$500k for the channel owner.***

Give a gift subscription

Humans can’t compete with the internet-clogging speed and scale that AI offers. I guess I’ll pop a subscribe button below to help siphon off some of the indignation:

Give a gift subscription

And now, back to work!

Workslop, that is.

You open a document from a colleague, it looks polished at first glance, but as you dig in, something’s off. It’s missing context. Some details smell a little off. The logic doesn’t quite connect. And then it hits you: someone just pointed an AI tool at the problem, copied whatever came out, and called it a day.

If this sounds familiar, congratulations; you’ve been workslopped.****

What Exactly Is Workslop?

The HBR article defines it beautifully: workslop is AI generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task.

The authors surveyed 1,150 US-based full-time workers, 40% of whom reported having received workslop in the last month. Among these recipients, how ugly is it on average?

• Average workslop share: 15.4% of workplace content.

• Primary vector: Mostly peer-to-peer (40%).

• Level of ambition: Workslop is ever so slightly more likely to rise (to managers, 18%) than to fall (from managers/execs 16%).

• Recipients felt: Annoyed (53%), confused (38%), offended (22%).

• Trust eroded: Ouch (see the chart in the next section).

• Invisible tax: 2h per instance, $186 per month per employee.

• Industries most impacted: Professional services and technology.

Equal parts titillating and unsurprising to anyone who understands that artificial intelligence is no match for natural stupidity.

Don’t get me wrong, I’m a firm believer in AI’s transformative potential… when we use it to reach higher. But when we use it to disengage, workslop is inevitable. Indeed, every kind of disengagement breeds its own kind of slop. You can do plenty of damage by disengaging about AI, not only with it.

What does workslop really represent? It’s thoughtlessness, enabled by AI.

Workslop happens when we let AI take over the thinking. When we treat it as a shortcut instead of a tool. When we optimize for looking productive instead of being productive.

Thoughtlessness enabled

AI is the great thoughtlessness enabler. It removes the friction that used to slow bad ideas and forced you to take your brain off screensaver mode to attempt something tricky. Until recently, getting a solution meant you sweated over the problem. Now you can skip wiggling the proverbial mouse and still produce something that looks polished to an equally checked out observer. This is true of both GenAI, which lets you vibe work*****, and traditional AI, which lets you turn garbage data you never bothered to check the provenance of into functional enterprise solutions. (Please tell me you didn’t need me to hang this massive SARCASM BANNER here.)

Unfortunately, thoughtlessness compounds. When no one double-checks, nonsense slips through as work. A single workslopped doc breeds more workslop. Your colleagues may mean well but the outputs they share may still be weak links riddled with problems hidden by polish. That creates a gap between good faith and reliable results — making the whole decision chain more fragile.

The cost of workslop isn’t just wasted time (an average of almost 2 hours lost per workslop instance in the HBR sample); it’s wrong turns taken with confidence.

According to HBR, every time you send workslop, half the people receiving it are downgrading their assessment of your competence. A third would prefer not to work with you in the future. It’s not just wasting time. It’s eroding trust. It’s damaging relationships. It’s degrading the quality of collaboration across your entire organization.

How to be the solution, not the problem

If AI enables thoughtlessness, it’s on us humans to re-inject thoughtfulness back into the process. Think of AI as a way to redistribute thoughtfulness towards your priorities, not as a way to skib** it.

That sounds simple, but simple and easy are not the same thing. AI makes thoughtlessness incredibly easy. That’s not going to change. So we need to make thoughtfulness an intentional practice.

Before you press that button, ask yourself:

• Value: Is this important? Does it matter if it’s wrong?

• Context: What specifically do I need here? What context is essential?

• Risk: What could go wrong if key details are missing?

• Check: What am I expecting to see? How would I verify this output?

If your task is important and your answer is “I won’t,” then stop right there.

After the AI gives you something, ask yourself:

• Does this actually meet my needs?

• Have I checked it enough?

• Would I stake my reputation on it?

Because you are. You absolutely are.

Leaders, your slop is so much grander

As more leaders hop on the “AI everywhere, all the time” bandwagon, brace yourself for… grandslop. Those AI mandates sound great in theory. In practice, they’re exponentially thoughtless.

You know what those dramatic ROI failure stats****** tell me about AI? That the problem is one of leadership, not technology. Leaders asleep at the wheel will use AI to burn budgets. Leaders with clear vision will use AI to define the next era of business.

For a proper dressing down, head over to HBR. I’ll keep it short here:

Wise leaders are those who model thoughtful AI use. Let your people see you thinking critically about AI outputs. Set clear guardrails — not to restrict, but to guide. Champion AI literacy and reward quality over velocity. I know that’s hard in a world that’s moving fast, but as they say: more haste, more slop.

Isn’t the irony rich?

Your organization adopted AI to save time. And now HBR is telling you to budget two of your precious hours towards cleaning up each instance of someone else’s careless AI use.

How about we all do our part to start turning that trend around? Next time you’re tempted to copy-paste some AI output directly into your deliverable, pause. Ask yourself: am I using this AI tool to reach higher? Or am I just being lazy? Am I creating work? Or am I creating workslop?

Because the world has enough slop already.

Thank you for reading — and sharing!

I’d be much obliged if you could share this post with the smartest leader you know.

Give a gift subscription

I recently taught a pilot of an everyone-friendly Decision-Making with ChatGPT course, so I’m seriously considering offering a 3h live virtual session (2h theory + 1h AMA) on Maven. Let me know if you’d be interested in taking it:

Yay for Cassie's New Course

In other news, the first few cohorts of my Agentic AI for Leaders course were a triumph and we’ve opened enrollment for an additional cohort to meet demand.

The course is specifically designed for business leaders, so if you know one who’d benefit from some straight talk on this underhyped overhyped topic, please send 'em this link: bit.ly/agenticcourse

🎤 MakeCassieTalk.com

Yup, that’s the URL for my public speaking. “makecassietalk.com” Couldn’t resist. 😂

Use this form to invite me to speak at your event, advise your leaders, or train your staff. Got AI mandates and not sure what to do about them? Let me help. I’ve been helping companies go AI-First for a long time, starting with Google in 2016. If your company wants the very best, invite me to visit you in person.

🦶Footnotes

* Concerned parents told Reddit that their toddler “refuses to use the toilet now due to skibidi toilet.** Whenever we try to put him on it he screams and refuses to go anywhere near it. We’ve tried explaining that Skibidi Toilet isn’t real and our toilet is completely safe, but it seems like it’s too overwhelming for him.”

** Skibidi toilet is an AI slop video (screenshot below to give you a sense of this High Art), that’s also the etymology of that slang your tween loves to irritate you with.

*** AI-generated slop is surging on YouTube... and its algorithm appears to favor AI over humans now. What makes bot inflation worth it for YouTube? AI lowers costs to cents per video while keeping quality “good enough.” If advertisers don’t object, YouTube may quietly let AI take over, leaving human creators sidelined until users suddenly notice: “there aren’t any humans on it anymore.” Your attention is valuable, so spend it wisely. The internet becomes what you make it.

**** HBR may have birthed the noun, but I… I shall raise the verb to be a fine participle.

***** Runner-up prize for second best neologism of the week goes to Microsoft, which just launched “vibe working” (its agent mode for Excel and Word). A gerund, how fancy!

****** 95% of organizations fail to see measurable ROI from GenAI. Details here.

🗞️ AI News Roundup!

In recent news:

1. Claude Sonnet 4.5 is here to work for 30 hours straight on complex coding tasks

Anthropic’s latest model works on complex tasks for over 30 hours — smoking the already-impressive 7h limit on OpenAI’s Codex — while setting records on real-world coding and computer-use tests. It also adds developer tools like checkpoints, code execution, and a VS Code extension, plus a new Agent SDK and stronger safety controls.

2. OpenAI debuts video generator Sora 2

Sora 2 makes a leap from Sora 1’s janky prompt-matched clips to physics-aware, persistent video with synced audio. The new iOS app (rolling out for US and Canada this week) adds cameo avatars, consent controls, and remix tools.

Meanwhile, Google DeepMind released a nifty paper showing that their video model Veo 3 can solve a wide range of visual tasks in a zero-shot manner, from perception and physics modeling to manipulation and reasoning, suggesting that video models are on track to become general-purpose foundation models for vision, much like LLMs transformed language understanding.

3. Accenture tells its staff to get AI literate or get lost

Accenture plans to lay off staff unable to reskill on AI as it doubles down on its restructuring, investing $865 million in a 6 month business optimization program. If your company is considering similar changes, there are humane ways and harsh ways to handle them. I hope you’ll pick the better option. Feeling lost? I’ve helped organizations build AI literacy and reskill workforces for over a decade, starting with Google’s AI-first shift. To get started, request a workshop at makecassietalk.com (flavors include board of directors / managers and execs / general staff — let me know what you need); it’ll cost you much, much less than $865 million.

4. Musk’s xAI undercuts rivals with federal deal to sell Grok chatbot for 42 cents per use

xAI secured an 18-month U.S. government contract letting executive agencies use Grok at less than half the $1 per-use price of OpenAI and Anthropic. The package includes integration support from xAI engineers, despite prior controversies over Grok generating offensive content. Some estimates put the cost of training Grok 4 at $490 million, underscoring the scale of investment behind the discounted entry into federal procurement.

5. Alibaba drops flurry of AI upgrades, including lip-reading live translator

Alibaba introduced Qwen3-Omni, its omni-modal model for text, images, audio, and video with real-time streaming, 119 language coverage, and benchmark wins over GPT-4o and Gemini-2.5 Pro. The launch also included Qwen3-VL (vision and language), Qwen3-Max (reasoning and coding), and Qwen3-LiveTranslate-Flash (translate with gesture recognition and lip-reading… in case you thought those distant cameras can’t hear you). Go big or go home!

6. Study suggests GenAI will affect tech jobs more than other sectors

An analysis by Indeed indicates tech roles face the deepest AI-driven changes: over half the skills in tech jobs could be transformed by generative AI (nearly 3 in 5 fully automatable skills are tech-related). Meanwhile, another report finds that 90% of software developers are using AI at work, with 65% “heavily reliant” on it.

7. MIT team teaches AI to design crystal patterns that could power better quantum computers

MIT and collaborators built a tool called SCIGEN that guides AI to create the rare crystal shapes most likely to unlock new quantum materials. The system generated over 10 million options, with supercomputer tests showing thousands could have unusual magnetic traits. Two new compounds were successfully made in the lab, pointing to a faster path toward more stable qubits.

Subscribe now

Forwarding this email to a friend or sharing it on social media goes a long way to encouraging my writing. Thank you thank you!