safety-research/petri
An alignment auditing agent capable of quickly exploring alignment hypothesis
2 best points · 2 comments · 3 Hacker News posts · 1,406 GitHub stars · View on GitHub
Hacker News discussions
- Anthropic's Petri (2 points, 2 comments, by kordlessagain, 2025-10-14)
- An alignment auditing agent capable of quickly exploring alignment hypothesis (2 points, 0 comments, by JnBrymn, 2025-10-16)
- Petri AI Testing 'Closes' possible solution without looking (2 points, 0 comments, by Utharian, 2025-11-07)