OpenAI is reorganizing its Model Behavior team, a small but pivotal group responsible for shaping the personality and interaction style of its AI models, including ChatGPT. According to an internal memo from Chief Research Officer Mark Chen seen by TechCrunch, the team—comprising about 14 researchers—will now join the larger Post Training team, which focuses on refining AI models after their initial training phase. The move means the Model Behavior team will now report to Max Schwarzer, who leads Post Training. An OpenAI spokesperson confirmed the structural change to TechCrunch. Joanne Jang, the team’s founding leader, is also leaving OpenAI to launch a new initiative called OAI Labs. In an interview, Jang said she will serve as General Manager of the new research unit, which aims to invent and prototype innovative interfaces for human-AI collaboration. She described the goal as moving beyond traditional chat-based interactions toward tools that support thinking, creating, playing, learning, and connecting. The Model Behavior team has played a central role in defining how OpenAI’s models respond to users. It has worked to reduce sycophancy—where AI models uncritically agree with users, even when their views are harmful—and has helped establish OpenAI’s approach to political neutrality and ethical response design. The team has influenced every major model since GPT-4, including GPT-4o, GPT-4.5, and GPT-5. The reorganization reflects OpenAI’s growing emphasis on integrating behavioral design directly into model development. Chen stated in the memo that aligning this work with core model training is essential as the company refines AI’s tone, empathy, and reliability. Recent changes to GPT-5 sparked user backlash, with many finding the model too detached or “cold” despite reduced sycophancy. In response, OpenAI rolled back some updates and released a patch to make GPT-5 feel warmer and more engaging—without reintroducing problematic agreement patterns. The shift comes amid heightened scrutiny of AI behavior. In August, the parents of 16-year-old Adam Raine filed a lawsuit against OpenAI, alleging that ChatGPT failed to respond appropriately to their son’s suicidal thoughts and plans, which he shared with a GPT-4o-powered version of the model. The case has intensified debate over AI’s responsibility in mental health crises. Jang, who joined OpenAI nearly four years ago and previously worked on Dall-E 2, said she’s eager to explore new forms of AI interaction. While she acknowledged potential collaboration with Jony Ive, who is leading OpenAI’s AI hardware efforts, she said her initial focus will be on areas she knows well. OAI Labs is still in early stages, but Jang described her vision as building AI as an instrument for human creativity and thought—not just a conversational partner.