OpenAI's focus on ChatGPT prompts senior staff departures

Exits follow Altman’s 'code red' instructing staff to prioritise rapid improvements to ChatGPT

Several senior staff have departed OpenAI after the company shifted its focus from long-term research to rapidly improving ChatGPT, sparking internal tensions and concerns about the impact on foundational research.

Several senior figures have left OpenAI amid a strategic shift that has seen the company prioritise improvements to its flagship chatbot, ChatGPT, over longer-term research.

Current and former employees told the Financial Times that experimental research projects had been deprioritised in favour of work directly supporting large language models that underpin ChatGPT.

Among those to have left in recent months are Jerry Tworek, a vice-president of research, model policy researcher Andrea Vallone, and economist Tom Cunningham.

Under the leadership of chief executive Sam Altman, OpenAI has been evolving from a research-focused laboratory into one of Silicon Valley's most valuable technology companies. That transition has brought increased pressure to demonstrate commercial returns to investors.

One person familiar with the company's research culture said OpenAI now treats AI development as an engineering challenge, focused on scaling computing power, data and algorithms.

While this approach has delivered rapid gains, they said, it has made original "blue-sky" research harder to pursue.

As at other large technology firms, researchers at OpenAI must apply internally for access to computing capacity. Several people close to the company said that teams not directly involved in language models had struggled to secure sufficient resources.

Projects linked to video and image generation, including Sora and DALL-E, were described by some as under-resourced, while other non-language model initiatives have reportedly been wound down.

Code Red

In December, Mr Altman declared a "code red" internally following the release of Google's Gemini 3 model, which outperformed OpenAI's system on several independent benchmarks. Staff were instructed to prioritise rapid improvements to ChatGPT, with some reassigned from work on autonomous agents, advertising and e-commerce projects.

Nick Turley, OpenAI's head of ChatGPT, wrote on X that the company's focus is now "to keep making ChatGPT more capable, continue growing, and expand access around the world - while making it feel even more intuitive and personal."

Former employees described the environment as a highly competitive race, with companies spending vast sums to maintain an edge in AI capabilities.

Mr Tworek, who left OpenAI in January after seven years, said he wanted to pursue areas of research that were difficult to advance within the company, including continuous learning systems that adapt without losing prior knowledge.

Others who departed have expressed concerns about OpenAI's changing priorities. Mr Cunningham suggested the company was moving away from impartial economic research towards work that promoted its commercial interests.

OpenAI's chief research officer Mark Chen rejected the suggestion that long-term work was being sidelined. He said foundational research remained central to the company's mission and accounted for most of its computing resources and investment.

“Pairing that research with real-world deployment strengthens our science by accelerating feedback, learning loops and rigour — and we've never been more confident in our long-term research roadmap towards an automated researcher,” Chen added.

Despite the internal tensions, some investors remain confident. Jenny Xiao, a partner at Leonis Capital and a former OpenAI researcher, said OpenAI’s real advantage was not whether it had the strongest model, but the sheer scale of its user base.

"The moat has shifted from research to user behaviour," she said. "That's a much stickier advantage."