Inspiration
We found our inspiration in three key areas: Childhood Memory: One of us recalled moments of accessing websites where messages popped up: "Free iPhone! Click here." We clicked, and nothing bad happened, but it could have. Daily News Headlines: Numerous stories highlight scams affecting not just kids, but adults too, the ideas drawn directly from reality. Our Gaming Experience We grew up learning complex games through safe failures. If we failed, we could always respawn. We wondered: Why can't online safety be learned the same way? That question became our mission.
What It Does
The Problem We Saw: Age between 8 to 12 years old kids are curious to everything they see, and parents are busy. When handed a phone or tablet, children enter a world full of strangers, scams, and hidden dangers. A "fun" link can lead to a phishing site. A "friendly" stranger might be a predator. Telling kids "be careful" isn't enough. They need to see the danger, experience it, and learn from it before it’s real.
Our Solution: We built SimShark - a sandboxed game that simulates the internet safely. Kids navigate fake versions of real apps: email, WhatsApp, Instagram, with 100% AI-generated content. They encounter:
- A scam email offering "free game coins" with a suspicious link.
- A "friendly" stranger in chat asking personal questions.
- A bully posted mean posts.
- A friend pressuring them to share a password.
At every choice, they decide what to do. Reply? Block? Tell an adult? Ignore? After each choice, our AI coach explains: "That person asked for your photo. In the real world, that could be dangerous. Here's why."
The Benefit: Kids don’t just hear rules, they build intuition, learn to feel the difference between trusted messages and scams, practice saying "no" to pressure, all in a consequence-free zone, preparing them for the real internet.
How We Built It
Time was of the essence, and it was the first time we were all working together, so we acted quickly.
Step 1: Foundation We huddled around iPads, analyzed the award topic, defined target users (kids aged 8-12), and wrote our problem statement.
Step 2: Divide and Conquer We split into three tracks of two: Track A: UI/UX Design
- Crafted a colorful, familiar interface that resembles real social media but feels safe for kids.
- Mapped the user journey: click in an app → face a scenario → make choices → get coached → repeat.
- Designed the coach characters
Track B: AI Prompt Engineering
- Tested how to prompt AI to play roles: a scammer, a bully, a concerned friend.
- Discovered AI restrictions and had to balance realism with safety.
- Built a library of prompts generating the right tone consistently.
*Track C: Backend & Integration *
- Handled architecture, connecting UI, AI engine, and game logic.
- Built the "scenario engine" based on children’s choices and fixed bugs.
Challenges We Ran Into
Insufficient time We had many ideas, but with limited hours, we focused on three core apps: Email, WhatsApp, and Social Media.
The AI Restriction Wall We wanted realistic AI characters, but every AI guardrail pushed back. Prompt engineers spent hours finding the right wording that triggered educational responses.
Integration Complexity Six people meant few pieces of code needing to fit together. We learned to check in constantly to ensure functionality before moving forward.
Team Communication With no time for awkward silences, we had to learn how to disagree quickly. When designers wanted features and coders said "no time," we found middle ground, prioritizing essential builds first.
Accomplishments We're Proud Of
We built a game in less than 12 hours. Not a prototype, but a working, interactive game for kids. From blank screen to playable demo in under half a day.
We let artificial intelligence (basically) act as intended. Despite some limitations, we let scammers commit fraud, bullies bully others, and coaches teach.
We built something that works. Within a day, we ran through a full scenario. It worked—kids could click, choose, and learn.
What We Learned
Choosing a topic is half the battle. We debated too long on which award to target. Next time, we’ll trust our gut faster.
Divide and conquer works. Splitting into pairs kept us accountable and made progress smoother amidst chaos.
Teamwork is a skill built in real time. We learned to communicate without ego, listen, and build trust quickly.
What's Next for SafeSims
If we had more time, we would build:
More Scenarios:
- Fake website ads too good to be true.
- Account setup simulations (password strength, privacy settings).
- Public WiFi security warnings.
Smarter AI:
- More accurate, nuanced characters.
- A wider variety of scams.
More Real-Life Simulations:
- Teaching kids not to share certain photos.
- Phone call scams.
- Friend pressure simulations.
- Location sharing warnings.
Log in or sign up for Devpost to join the conversation.