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