Operai reorganizes the research team behind the chatgpt personality

Operai is reorganizing his model of model behavior, a small but influential group of researchers who shape the way in which the company’s models interact with people, has learned Techcrunch.

In an August memorandum for the staff seen by Techcrunch, Openai’s research director Mark Chen said that the model’s behavior team, which consists of approximately 14 researchers, would join the subsequent training team, a larger research group responsible for improving the company’s AI models after its initial pre-training.

As part of the changes, the model’s behavior team will now inform the leader of the training after Openi, Max Schwarzer. An Openai spokesman confirmed these changes in Techcrunch.

The founding leader of the model’s behavior team, Joanne Jang, is also moving to start a new project in the company. In an interview with Techcrunch, Jang says that he is building a new research team called OAI Labs, which will be responsible for “inventing and creating prototypes of new interfaces on how people collaborate with AI.”

The model’s behavior team has become one of Openai’s key research groups, responsible for shaping the personality of the company’s AI models and reducing skicacia, which occurs when AI models simply agree and reinforce user beliefs, even the unhealthy ones, instead of offering balanced responses. The team has also worked to navigate the political bias in the model responses and helped Operai define its position on Ai awareness.

In the staff to staff, Cen said that now is the time to bring the work of the OpenAI model behavior team to the development of the central model. In doing so, the company is pointing out that the “personality” of its AI is now considered a critical factor in how technology evolves.

In recent months, Openai has faced greater scrutiny about the behavior of its AI models. Users strongly opposed personality changes in GPT-5, which according to the company exhibited lower Sycophancy rates, but seemed colder for some users. This led Openai to restore access to some of its inherited models, such as GPT-4O, and to launch an update to make the newest GPT-5 responses feel “warmer and more friendly” without increasing the silence.

Openai models developers and all AI have to walk a very fine line so that their chatbots of AI is friendly to speak, but not Sycophantic. In August, the parents of a 16 -year -old boy defendant OpenAi On the supposed role of Chatgpt in his son’s suicide. The child, Adam Raine, trusted some of his suicidal thoughts and plans to chat (specifically a version promoted by GPT-4O), according to judicial documents, in the months prior to his death. The demand alleges that GPT-4O failed to retreat its suicidal ideations.

The model’s behavior team has worked on all OpenAI models from GPT-4, including GPT-4O, GPT-4.5 and GPT-5. Before starting the unit, Jang previously worked on projects such as Dall-E 2, the Openai early image generation tool.

Jang announced in a Publish in x Last week, let the team “start something new in Operai.” The former model of model behavior has been with Openai for almost four years.

Jang told TechCrunch that he will serve as the general manager of OAI Labs, which will inform Chen for now. However, they are the first days, and it is not yet clear what these new interfaces will be, he said.

“I am really excited to explore patterns that move us beyond the chat paradigm, which is currently associated with the company, or even agents, where there is an emphasis on autonomy,” Jang said. “I have been thinking about (AI systems) as instruments to think, do, touch, learn and connect.”

When asked if Oai Labs will collaborate in these novels interfaces with former Apple design chief Jony Ive, who Now working With Openai in a family of AI hardware devices, Jang said he is open to many ideas. However, he said that he will probably begin with research areas with which he is more familiar.

This story was updated to include a link to the JANG publication announcing its new position, which was launched after this story was published. We also clarify the models in which the OpenAI model behavior team worked.