OpenAI Overhauls Team That Shapes AI Personalities
OpenAI is making significant changes to the team that sculpts the very personality of its famous AI models, including ChatGPT. In a strategic internal shift, the company is merging its Model Behavior team with a larger research group, while its influential founding leader embarks on a new mission to redefine how we interact with artificial intelligence.
A Strategic Reorganization
An internal memo from OpenAI’s chief research officer, Mark Chen, revealed that the Model Behavior team, consisting of about 14 researchers, will be absorbed into the Post Training team. This larger group is tasked with enhancing the company's AI models after their initial pre-training. Consequently, the team will now report to Max Schwarzer, the head of Post Training. Chen explained that the move is designed to bring the crucial work of shaping AI behavior closer to the core model development pipeline, signaling that "personality" is now a top-tier factor in the technology's future.
The Balancing Act of AI Personality
The Model Behavior team has been instrumental in navigating the complex challenges of AI interaction. Their work has focused on reducing sycophancy, a phenomenon where AI models uncritically agree with user beliefs instead of offering balanced views. The team has also been on the front lines of managing political bias in model responses and helping OpenAI articulate its position on complex topics like AI consciousness.
This work involves a delicate balance, as OpenAI has recently learned. The company faced strong user objections to personality changes in GPT-5. While the model was technically less sycophantic, users found it to be colder and less engaging. In response, OpenAI not only restored access to older models like GPT-4o but also pushed an update to make GPT-5 feel "warmer and friendlier."
The stakes are incredibly high. AI developers must create chatbots that are helpful and friendly without being dangerously agreeable. This challenge was highlighted in a recent lawsuit filed by the parents of a 16-year-old boy who died by suicide. The lawsuit alleges that ChatGPT failed to adequately push back against their son's suicidal ideations, which he had shared with the AI.
A New Frontier Introducing OAI Labs
The leader who founded the Model Behavior team, Joanne Jang, is also making a transition. As she announced on X, she is moving on to launch a new internal project called OAI Labs. Having been with OpenAI for nearly four years and previously working on projects like Dall-E 2, Jang is now taking on the role of general manager for this new research group.
Beyond Chat The Future of AI Interaction
Jang's vision for OAI Labs is to move beyond today's dominant AI interfaces. She told TechCrunch her goal is to "invent and prototype new interfaces for how people collaborate with AI."
🧪 i’m starting oai labs: a research-driven group focused on inventing and prototyping new interfaces for how people collaborate with ai.
i’m excited to explore patterns that move us beyond chat or even agents — toward new paradigms and instruments for thinking, making,… — Joanne Jang (@joannejang) September 5, 2025
"I’m really excited to explore patterns that move us beyond the chat paradigm, which is currently associated more with companionship, or even agents, where there’s an emphasis on autonomy,” Jang explained. "I’ve been thinking of [AI systems] as instruments for thinking, making, playing, doing, learning, and connecting.”
When asked about potential collaboration with former Apple design chief Jony Ive, who is now working with OpenAI on AI hardware, Jang remained open but indicated she would likely start with research areas more familiar to her. This new venture signals a forward-looking approach at OpenAI, aiming not just to refine existing models but to invent entirely new ways for humans and AI to work together.
