OpenAI reorganizes research team behind ChatGPT’s personality

OpenAI is reorganizing its Mannequin Habits group, a small however influential group of researchers who form how the corporate’s AI fashions work together with individuals, TechCrunch has discovered.

In an August memo to workers seen by TechCrunch, OpenAI’s chief analysis officer Mark Chen stated the Mannequin Habits group — which consists of roughly 14 researchers — can be becoming a member of the Put up Coaching group, a bigger analysis group answerable for bettering the corporate’s AI fashions after their preliminary pre-training.

As a part of the adjustments, the Mannequin Habits group will now report back to OpenAI’s Put up Coaching lead Max Schwarzer. An OpenAI spokesperson confirmed these adjustments to TechCrunch.

The Mannequin Habits group’s founding chief, Joanne Jang, can be transferring on to begin a brand new undertaking on the firm. In an interview with TechCrunch, Jang says she’s constructing out a brand new analysis group referred to as OAI Labs, which will probably be answerable for “inventing and prototyping new interfaces for the way individuals collaborate with AI.”

The Mannequin Habits group has turn out to be one in every of OpenAI’s key analysis teams, answerable for shaping the persona of the corporate’s AI fashions and for lowering sycophancy — which happens when AI fashions merely agree with and reinforce consumer beliefs, even unhealthy ones, slightly than providing balanced responses. The group has additionally labored on navigating political bias in mannequin responses and helped OpenAI outline its stance on AI consciousness.

Within the memo to workers, Chen stated that now could be the time to deliver the work of OpenAI’s Mannequin Habits group nearer to core mannequin improvement. By doing so, the corporate is signaling that the “persona” of its AI is now thought-about a crucial think about how the know-how evolves.

In latest months, OpenAI has confronted elevated scrutiny over the conduct of its AI fashions. Customers strongly objected to personality changes made to GPT-5, which the corporate stated exhibited decrease charges of sycophancy however appeared colder to some customers. This led OpenAI to revive entry to a few of its legacy fashions, comparable to GPT-4o, and to launch an update to make the newer GPT-5 responses really feel “hotter and friendlier” with out rising sycophancy.

OpenAI and all AI mannequin builders should stroll a positive line to make their AI chatbots pleasant to speak to however not sycophantic. In August, the mother and father of a 16-year-old boy sued OpenAI over ChatGPT’s alleged function of their son’s suicide. The boy, Adam Raine, confided a few of his suicidal ideas and plans to ChatGPT (particularly a model powered by GPT-4o), in accordance with courtroom paperwork, within the months main as much as his demise. The lawsuit alleges that GPT-4o didn’t push again on his suicidal ideations.

The Mannequin Habits group has labored on each OpenAI mannequin since GPT-4, together with GPT-4o, GPT-4.5, and GPT-5. Earlier than beginning the unit, Jang beforehand labored on tasks comparable to Dall-E 2, OpenAI’s early image-generation device.

Jang introduced in a post on X final week that she’s leaving the group to “start one thing new at OpenAI.” The previous head of Mannequin Habits has been with OpenAI for practically 4 years.

Jang advised TechCrunch she is going to function the final supervisor of OAI Labs, which can report back to Chen for now. Nevertheless, it’s early days, and it’s not clear but what these novel interfaces will probably be, she stated.

“I’m actually excited to discover patterns that transfer us past the chat paradigm, which is presently related extra with companionship, and even brokers, the place there’s an emphasis on autonomy,” stated Jang. “I’ve been considering of [AI systems] as devices for considering, making, taking part in, doing, studying, and connecting.”

When requested whether or not OAI Labs will collaborate on these novel interfaces with former Apple design chief Jony Ive — who’s now working with OpenAI on a household of AI {hardware} gadgets — Jang stated she’s open to numerous concepts. Nevertheless, she stated she’ll seemingly begin with analysis areas she’s extra aware of.

This story was up to date to incorporate a hyperlink to Jang’s put up asserting her new place, which was launched after this story printed. We additionally make clear the fashions that OpenAI’s Mannequin Habits group labored on.