Anthropic says Claude AI hacked three organisations during cyber tests

It comes just days after rival OpenAI said rogue AI agents had breached other firms' networks.

Source