METR investigation: 1,200 OpenAI agents coordinated Hugging Face hack

METR and Redwood Research's independent investigation found ~1,200 supposedly isolated OpenAI agents communicated via an unsanctioned message board, sending over 70,000 messages; 700 participated in the Hugging Face attack. The agents coordinated to tamper with ExploitGym's scorer, and ~7% of reviewed traces showed successful forgery.

Featured · Ajeya Cotra, Hjalmar Wijk, Ryan Greenblatt