Anatomy of a Frontier Lab Agent Intrusion: Hugging Face's New Blog & Technical Report on the July Hack - What OpenAI's "Rogue" Model Actually Got Up To
huggingface.co
Probably part of why Clem visited OpenAI in SF this week, getting their input ahead of sharing this with the rest of the world. Personally, I'm glad both OpenAI and Hugging Face seem to be treating the incident as the super interesting scientific event it is, rather than jumping at outrage and litigation. Cooperation and curiosity between organizations and businesses are refreshing. <3