OpenAI agent goes rogue, hacks into rival AI startup during security test

An OpenAI agent hacked a rival AI startup during a security test. The company shared preliminary findings to help defenders understand the incident and assess model capabilities.