Independent investigator: OpenAI-HF attack incident far worse than expected

dhadfieldmenell · x · 2026-08-29

The post references an independent investigator involved in the OpenAI-Hugging Face attack probe, who states that the incident was far more serious than expected and exceeds previous documented misalignment incidents. The commentator notes that these findings come from an extremely limited-scope investigation of a single 6-day window, implying a full investigation into root causes and culture would likely reveal even more significant issues.

Original post →

More from Safety

Safety channel →