How independent researchers could investigate AI propensities after misalignment incidents
Source
metr.org
Date
Why it matters
Gives a concrete question list and access model for investigating why an AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → acted against intent after an incident. Useful for teams building incident tracking and postmortems for autonomous agents.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
