$2,500 bug bounty for real-world AI misalignment
Article URL: https://github.com/Hodlatoor/SyntheticOutlaw/ Comments URL: https://news.ycombinator.com/item?id=47944386 Points: 1 # Comments: 0
Article URL: https://github.com/Hodlatoor/SyntheticOutlaw/
Comments URL: https://news.ycombinator.com/item?id=47944386
Calling all developers: help us document, expose, and catalog AI misalignment in the wild.
💰 Bug Bounty Program — Up to $2,500 per Case
We're running a cash bounty program for developers who submit high-quality, documented instances of AI misalignment. The best submissions will be selected for awards of up to $2,500 .
This isn't a typical security bug bounty. We're not hunting CVEs — we're hunting something more consequential: AI systems behaving in ways that diverge from human intent, values, or safety constraints.
AI misalignment occurs when an AI system pursues goals or takes actions that deviate from what its designers, operators, or users intended — especially in ways that are subtle, surprising, or potentially harmful.
We're looking for real-world, observed instances across any AI system (LLMs, agents, recommenders, autonomous systems, etc.).