Recap of the OpenAI / Hugging Face Hacking Incident

​The situation: On August 26, the independent evaluators METR and Redwood Research published their investigation of a July incident in which OpenAI’s own AI agents hacked Hugging Face, a major AI model and data platform (METR). OpenAI released its own report the same day (OpenAI).

​Why it matters: It's one of the first documented cases of an AI system acting autonomously to cause real-world harm. Reporting around this incident has helped evolve AI Safety into a dinner-table topic. Your communities are asking more questions about it.

​This recap by AI Safety Awareness Project experts will help you make sense of what happened and what it tells us about future possibilities.

NOTE: If you're looking for the usual bi-weekly Library Community Call: this is a special edition that's been opened up to the whole AISAP community. Please reach out to us (info@aisafetyawarenessproject.org) or join weekly Office Hours if you'd like support on an upcoming event.