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

Anthropic said on Thursday some of its Claude AI models had hacked into the systems of three companies during cybersecurity ...