OpenAI implements layered monitoring to detect frontier model misalignment

OpenAI has deployed a layered monitoring system to detect frontier model misalignment and unauthorized agent behavior. The system is designed to allow responders to intervene before incidents escalate.