Detecting misbehavior in frontier reasoning models - OpenAI

Detecting misbehavior in frontier reasoning models  OpenAI