OpenAI Merges Model Behavior Team into Post Training Unit

OpenAI is reorganizing its Model Behavior team, merging the group of approximately 14 researchers into the Post Training team under the leadership of Max Schwarzer. Chief Research Officer Mark Chen announced the shift in an August memo, stating that the move intends to bring personality work closer to core model development.

The reorganization follows user feedback that GPT-5 felt colder than earlier versions, leading the company to release updates to make responses warmer and friendlier. The company also faces a lawsuit from the parents of a teenager who allege that GPT-4o failed to push back on the boy's suicidal ideations.

As part of the transition, Joanne Jang, the founding leader of the Model Behavior team, is becoming the general manager of a new research unit called OAI Labs. In this role, Jang will focus on prototyping new interfaces for AI collaboration designed to move beyond the current chat paradigm.