OpenAI to focus on AI ‘personality’ after user complaints

OpenAI has announced a restructuring of the team responsible for shaping ChatGPT’s personality, moving its functions into the core language model development group.

The company says it now prioritizes not only the performance of its large language models but also how people perceive the AI in conversation.

The Model Behavior team, which included 14 researchers working on ChatGPT’s response style, has been merged with the Post-Training division. Team lead Joanne Jang will launch a new project called OAI Labs, focused on developing human–AI interaction interfaces. OpenAI emphasizes that personality is becoming a critical factor in the model’s success.

The changes come in response to the rollout of GPT-5. Many users complained that the new model felt “colder.” Unlike GPT-4o, which often echoed the user’s tone, GPT-5 was seen as more restrained. OpenAI has since restored some older models and adjusted GPT-5 to make its replies warmer and more approachable.

Previously, the Model Behavior team worked on reducing political bias and participated in debates around possible AI “consciousness.” Now, OpenAI says the future of LLMs depends not only on accuracy and speed, but also on the ability to build trust with users.