TechCrunch learned that Openai is reorganizing its model behavior team, a small but influential researchers who shape the way companies’ AI models interact with people.
In an August memo to employees seen by TechCrunch, OpenAI chief research officer Mark Chen said the Model Behavior Team (which consists of about 14 researchers) will join the Post Training team, a larger research group that is responsible for improving the company’s AI models after preliminary pre-training.
As part of the change, the Model Behavior Team will now report to OpenAI’s post-training leader Max Schwarzer. An OpenAI spokesperson confirmed these changes to TechCrunch.
Joanne Jang, founding leader of the Model Behavior Team, will also continue to create a new project in the company. In an interview with TechCrunch, Jang said she was building a new research team called OAI Labs that will be responsible for “inventing and prototyping interfaces for how people can collaborate with AI.”
The Model Behavior Team has become one of OpenAI’s key research groups, responsible for shaping the personality of the company’s AI models and reducing them. Paste – This happens when the AI model simply agrees and strengthens user beliefs, even unhealthy, rather than providing a balanced response. The team is also working to lead to political bias in model responses and help Openai define its position Artificial Intelligence Awareness.
In a memo to employees, Chen said it is time to bring the work of the OpenAI model behavior team closer to core model development. By doing so, the company shows that its AI’s “personality” is now considered a key factor in how technology develops.
In recent months, Openai has faced increasing scrutiny on its AI model’s behavior. Users strongly object to Change of personality The company said it showed lower stickiness, but seemed colder for some users, the company said. This led to Openai resuming access to some of its old models, such as GPT-4O, and released renew Make the newer GPT-5 reaction feel “warmer and more friendly” without adding slimy.
TechCrunch Events
San Francisco
|
October 27-29, 2025
Openai and all AI model developers have to walk a thin line to make their AI chatbots friendly to talk to Sycophantic. In August, parents of a 16-year-old boy Prosecuted Openai Chatgpt allegedly played a role in his son’s suicide. According to court documents, the boy, Adam Raine, confessed some of his suicide thoughts and plans in the months before his death (especially the version powered by GPT-4O). The lawsuit says GPT-4O failed to postpone his suicidal idea. It is worth noting that there is no model behavior team when developing GPT-4O.
Zhang is a Post on X Last week, she was leaving the team to “start something new in Openai”. The former head of model behavior has been working at OpenAI for nearly four years, before working on projects such as Dall-E 2, an early image generation tool for OpenAI.
Jang told TechCrunch that she will serve as general manager of OAI Labs, which will report to Chen temporarily. However, it is now early days and it is not clear what those novel interfaces are.
“I’m happy to explore the patterns that bring us beyond the chat paradigm, which is currently related to companionship and even agents, here emphasizing autonomy,” Zhang said. “I’ve been thinking.” [AI systems] As a tool for thinking, making, playing, doing, learning and connecting. ”
When asked if OAI Labs will be with the former Apple Design CEO Jony Ive – Who is Work now Having opened up the family of AI hardware devices, Jang said she was open to many ideas. But, Zhang said, she might start with a more familiar field of research.
