How we detect frontier model misalignment
📝
内容提要
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.
🏷️
