Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident

metr.org

saved by: 0 people
related read on: Astra and Fable still hack on simple variants of alignmen... from the discussion

METR and Redwood Offer Holy #%^@ Postmortem Of The Huggin... from the discussion

OpenAI halts training of latest models as reports mount o... from the discussion

Is A.I. Above the Law? from the discussion

Revealing the details of how OpenAI agents hacked Hugging... from the discussion