Related

Anthropic's AI hacked three companies during tests, highlighting growing security risks