OpenAI Reorganizes Research Teams, Merges Model Behavior with Post Training

OpenAI has restructured its research teams, merging the Model Behavior group with the Post Training team. The move, detailed in an August memo from Chief Research Officer Mak Chen, directs the 14-member Model Behavior team to join Post Training, which focuses on refining AI models after initial pre-training. Model Behavior will now report to Post Training lead Max Schwarzer.

Joanne Jang, founding leader of the Model Behavior team, is departing to launch OAI Labs, a new research initiative aimed at prototyping innovative human-AI collaboration interfaces. The Model Behavior team has played a pivotal role in shaping OpenAI's AI personality, reducing sycophancy in model responses, and addressing political bias—critical work that will now continue under the Post Training umbrella.