OpenAI senior staff members have departed as the company redirects resources from long-term research to its ChatGPT product, according to a Financial Times report citing 10 current and former employees.
Vice President of Research Jerry Tworek, who led development of OpenAI’s o1 reasoning model, left in early January 2025 after nearly seven years with the company. In an internal memo shared on X, Tworek stated he was “leaving to try and explore types of research that are hard to do at OpenAI.” His exit reflects the challenges some researchers face in pursuing certain projects within the current organizational structure.
Model Policy Researcher Andrea Vallone, who headed safety research on how ChatGPT responds to users experiencing mental-health distress, has joined rival Anthropic. Her work focused on ensuring appropriate safeguards in the chatbot’s interactions during sensitive user situations.
Economist Tom Cunningham departed in September 2025. In an internal farewell message, he cited a “growing tension between conducting rigorous analysis and serving as an informal promotional entity for OpenAI.” Cunningham’s role involved economic assessments that sometimes conflicted with the company’s public positioning.
Researchers on non-large language model projects have encountered denied or scaled-back funding requests. Teams responsible for Sora, the video generation tool, and DALL-E, the image generation model, operate with limited resources. Several non-language projects have been wound down as part of a broader reorganization at OpenAI.
Chief Research Officer Mark Chen addressed these changes in comments to the Financial Times. He affirmed that OpenAI remains committed to long-term foundational research. Chen explained that combining research with real-world deployment “strengthens the company’s science by speeding up feedback, learning cycles, and rigor.” Following Tworek’s departure, Chen stated in a company message that OpenAI is “very excited about our 2026 roadmap and advancing work toward an automated scientist.”
The strategic pivot originated from a “code red” memo by CEO Sam Altman in December 2025. The memo directed efforts to improve ChatGPT amid Google’s Gemini 3 receiving widespread acclaim and Anthropic expanding its enterprise client base. ChatGPT now serves over 800 million weekly users, while OpenAI contends with investor expectations for revenue that matches its valuation.
OpenAI’s trajectory has evolved from a research laboratory into a product-driven technology company. The organization prepares for a fourth-quarter initial public offering and is expanding its finance team to precede Anthropic in market entry. Altman has indicated that OpenAI can “dramatically” slow hiring while boosting output, enabled by AI to accomplish more without rapid workforce growth. In 2025, the company achieved an annual revenue run-rate exceeding $20 billion.
