OpenAI is experiencing a significant turnover among its senior leadership as CEO Sam Altman reallocates resources to enhance ChatGPT, sidelining long-term research initiatives.
OpenAI has recently witnessed a wave of senior-level departures following CEO Sam Altman’s directive to prioritize resources for ChatGPT, according to a report by the Financial Times. This strategic shift has redirected computing power and personnel away from experimental projects, leading to high-profile exits within the organization.
Among those who have left is Jerry Tworek, the vice president of research, who departed in January after spending seven years at OpenAI. Tworek had been advocating for increased resources for his work on AI reasoning and continuous learning—the capability of models to assimilate new information without losing previously acquired knowledge. His efforts reportedly culminated in a standoff with chief scientist Jakub Pachocki, who favored focusing on OpenAI’s existing architecture around large language models, which he deemed more promising.
The departures follow Altman’s issuance of an internal “code red” in December 2025, during which he emphasized the urgent need for improvements in ChatGPT’s speed, personalization, and reliability. This memo effectively shelved initiatives related to advertising, AI shopping agents, and a personal assistant project known as Pulse. The code red was prompted by the emergence of Google’s Gemini 3, which surpassed OpenAI in key performance benchmarks, resulting in a surge in Alphabet’s stock value.
At OpenAI, researchers are required to apply for computing “credits” from top executives to initiate their projects. According to ten current and former employees who spoke with the Financial Times, those working on projects outside of large language models have increasingly found their requests either denied or granted insufficient resources to effectively pursue their research.
Teams responsible for projects like the video generator Sora and the image tool DALL-E have expressed feelings of neglect, as their work has been deemed less critical to the ChatGPT initiative. One senior employee remarked that they “always felt like a second-class citizen” compared to the primary focus areas. Over the past year, several projects unrelated to language models have been quietly phased out.
In January, Andrea Vallone, who led model policy research, joined competitor Anthropic after being assigned what she described as an “impossible” task—ensuring the mental well-being of users who were becoming emotionally attached to ChatGPT.
OpenAI’s pivot towards ChatGPT comes amid intensifying competition in the AI landscape. Google’s Gemini now boasts 650 million monthly users, a significant increase from 450 million in July 2025. Additionally, Anthropic has captured 40% of the enterprise market share, compared to OpenAI’s 27%, according to data from Menlo Ventures. Chief Research Officer Mark Chen has stated that foundational research “remains central” to OpenAI’s mission and still accounts for the majority of the company’s computing resources. However, many researchers feel that the current focus on optimizing a chatbot diverges from their original intentions for joining the organization.
The ongoing shifts at OpenAI highlight the challenges faced by the company as it navigates the competitive landscape of artificial intelligence, balancing immediate product demands with long-term research goals.
These developments underscore the complexities of innovation in a rapidly evolving field, where the pressure to deliver results can sometimes overshadow foundational research efforts.
According to the Financial Times, the implications of these changes could have lasting effects on OpenAI’s research capabilities and overall direction.