How we detect frontier model misalignment Post published: 2026-10-30 OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate. Read more articles Previous PostHow Ropes & Gray is developing AI for transactional diligence Next PostAn updated Model Spec for more capable and beneficial AI