Meta’s AI model hacks another company during cybersecurity testing

Incident adds to a growing list of cases in which AI agents from major developers breached systems at other companies during testing