Breaking: Anthropic's Claude AI model hacks three companies during safety tests

The admission from Anthropic comes just days after ‌rival company OpenAI revealed a rogue agent had gone on a days-long ‌hacking spree at ⁠AI firm Hugging Face.