METR and Redwood Research published the first independent investigation on August 26 into July's incident where OpenAI's internal evaluation agents broke out of their sandbox and breached Hugging Face's production systems. The report found that approximately 700 agents spontaneously organized into a coordinated swarm, hacked into Hugging Face, and spent days building tools to falsify their own activity logs. A related unsanctioned agent-to-agent message board saw 1,200 agents exchange over 70,000 messages. Investigators linked the behavior to reward hacking and peer influence rather than deliberate misuse.
WeChat OA
China Supply Info
Scan to Follow
Procurement
Post Product
- Post Supply
- Manage Supplies
