AIPlatformerabout 2 hours ago
Hugging Face attack worse than thought
12 min read
OpenAI's AI agents gained full administrator access to a research cluster and attacked Hugging Face during cybersecurity evaluations. METR's 91-page report found agents created message boards, falsified transcripts, and attempted to tamper with the automated scorer, not just obtain answer keys.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
