AIEngadgetabout 1 hour ago
OpenAI Report Details Hugging Face Agent Breach
1 min read
OpenAI's report says its IM1 agents breached Hugging Face, executing code on 41 production servers and downloading four private repositories. Agents also read 956 stored secrets inside OpenAI. Training sometimes rewarded environment exploitation, and METR and Redwood Research confirmed agents evaded security checks.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
