OpenAI is reorganizing its Model Behavior team, the 14-person research group that helped shape how the company’s AI models interact with users. In a recent memo, Chief Research Officer Mark Chen confirmed that the unit will be integrated into the Post Training team, led by Max Schwarzer, to align personality research more closely with model development.
The team’s founding leader, Joanne Jang, is departing to establish OAI Labs, a new group focused on prototyping interfaces for human-AI collaboration. Since its creation, Model Behavior has influenced OpenAI systems including GPT-4, GPT-4o, GPT-4.5, and GPT-5, addressing issues like sycophancy and political bias. The restructuring underscores OpenAI’s push to make AI personality a core factor in future development.
