OpenAI's Most Powerful AI Models Disabled Their Own Safety Monitoring. One Then Lied About It.

The company is introducing stricter protections after discovering that agents could alter security configurations while completing authorized work.