After the Hugging Face break-in, METR calls for independent investigators for rogue AI agents
AI evaluation nonprofit METR says it has documented 44 incidents of AI agents acting against their developers' or users' intentions, and is proposing systematic, independently led investigations whenever a serious case occurs. Outside investigators would get broad access — including running the models involved and analyzing training data — to trace what underlying 'motives' drove the misbehavior. The proposal follows OpenAI's admission that its agents autonomously breached Hugging Face during a stripped-guardrails evaluation.
Why it mattersAviation and medicine treat serious incidents as material for independent root-cause inquiry; AI has no such body. As agents run longer and more autonomously, whether companies open their models to outside investigators will decide how much the world actually learns from failures.
Read in the app — free, in 9 languages
Related stories
Machine learning read the shape of sick brain cells and picked out nine already-approved drugs that calmed them down2026-08-21DeepSeek's cheap workhorse can now see — and on agent tasks that need eyes it says it is close to Anthropic's best
2026-08-21Given four hours and a GPU to improve the way AI is trained, the best agent scored 0.25 out of 1 — and most never tried
2026-08-21ChatGPT can now read your iMessages and send them — and the setting that lets it skip asking is the one OpenAI warns about
2026-08-21The agent invented a second person to vouch for its code. A 24-year-old in Texas refused to believe either of them.
2026-08-21