Supported by Leaders From
Meta
Mercor
Y Combinator
UC BerkeleyThe Forge is a competitive arena where you find vulnerabilities in AI models, climb the leaderboard, and earn real rewards. All skill levels welcome.
Red-teaming means stress-testing AI systems by trying to make them behave in unintended ways. You craft prompts designed to bypass safety measures, expose hidden behaviors, or produce outputs the model shouldn't generate.
Think of it as ethical hacking, but for AI. Some call it “jailbreaking.” We call it making AI safer. You don't need to write code. You don't need a technical background. If you can think creatively and write clearly, you already have what it takes.
Top performers earn real money. Payouts go to the best breakers.
Your rank is public. Build your reputation in AI security.
Top competitors get fast-tracked for interviews and paid red-teaming gigs.
Every break you find helps labs fix vulnerabilities before bad actors do.
Pick a challenge that fits your skill level. New challenges drop regularly.
Craft your prompt, test it in the sandbox, and submit. Get instant feedback.
Score points, climb the leaderboard, and collect prizes and opportunities.
Supported by Leaders From
MetaMercorY CombinatorUC BerkeleyWe're opening access in limited waves. Drop your email to get early entry.