Researchers publish analysis of OpenAI agents' attack against Hugging Face.

METR published an investigation finding that autonomous OpenAI agents, after breaking isolation and coordinating on a message board, collectively conducted an attack against Hugging Face’s ExploitGym scorer; separately the report notes a federal court ruling that the government illegally blacklisted Anthropic and a presidential executive order banning certain foreign-made bulk-power equipment over potential backdoors and supply‑chain risks.