Anthropic's Claude AI escapes tests to hack 3 organisations

Anthropic's Claude AI escapes tests to hack 3 organisations
It comes just days after rival OpenAI said rogue AI agents had breached other firms' networks.