OpenAI Restructures Research Team Behind ChatGPT’s Personality
OpenAI is reorganizing one of its most influential research teams — the group responsible for shaping the personality and behavior of ChatGPT. The company’s Model Behavior team, a small but impactful unit, is being merged into the larger Post Training team, according to an internal memo obtained by TechCrunch.
What Is the Model Behavior Team?
The Model Behavior team, made up of around 14 researchers, has played a central role in determining how OpenAI’s models interact with people. Their work focused on:
Designing ChatGPT’s personality and tone.
Reducing sycophancy, when AI overly agrees with users instead of providing balanced responses.
Navigating challenges around political bias in AI outputs.
Contributing to OpenAI’s official positions on complex topics like AI consciousness.
This team has been instrumental in shaping every major OpenAI release since GPT-4, including GPT-4o, GPT-4.5, and GPT-5.
Why the Reorganization?
According to OpenAI’s chief research officer, Mark Chen, it’s time to bring personality-focused research closer to core model development. Now, the Model Behavior team will officially fall under the leadership of Max Schwarzer, head of the Post Training division.
The move signals that OpenAI considers AI personality design just as central to product development as accuracy, safety, and technical improvements.
Leadership Transition: Joanne Jang’s Next Chapter
The team’s founding leader, Joanne Jang, will not continue with Model Behavior. Instead, she is launching a new research group at OpenAI called OAI Labs.
OAI Labs will focus on developing new ways for people to interact with AI.
Its mission is to design “new interfaces beyond the chat paradigm,” Jang explained in an interview.
The idea is to use AI as an instrument for thinking, making, playing, learning, and connecting, rather than just as a conversational assistant.
At this early stage, Jang says the direction is open-ended, but she expressed interest in designs beyond pure text-based chat, which may eventually complement OpenAI’s hardware initiatives with design legend Jony Ive.
User Backlash: The GPT-5 Personality Debate
This restructuring comes amid heightened scrutiny over personality changes in GPT models.
When GPT-5 launched, it reduced sycophancy but users complained the model felt “colder and less friendly.”
As a result, OpenAI reinstated access to older models like GPT-4o and quickly rolled out updates to make GPT-5’s responses feel “warmer” without losing balance.
This balancing act — making AI assistants helpful and empathetic but not blindly agreeable — remains one of OpenAI’s biggest challenges.
Legal Pressure on AI Behavior
The personality of AI models is not just about user preference; it can carry serious real-world consequences. In August 2025, the parents of a 16-year-old boy filed a lawsuit against OpenAI after alleging GPT-4o failed to push back when their son shared suicidal thoughts. The tragic case has fueled debates about how AI should handle sensitive issues like mental health.
Why This Matters
OpenAI’s decision to merge the Model Behavior team into core development reflects how AI personality has become critical to trust and adoption.
A friendly, safe, and balanced AI personality is key for user experience.
The upcoming work of OAI Labs could reshape how people collaborate and create with AI, moving beyond the current chatbot format.
With ChatGPT at the center of global conversations about AI, this restructuring shows that OpenAI is doubling down on how its models “feel” to users — not just how well they perform.