John Schulman, a founding member of OpenAI, has left the company to join rival AI startup Anthropic. Schulman, who played a pivotal role in developing OpenAI’s AI alignment research and the reinforcement training of models like ChatGPT, announced his decision to transition to Anthropic via a post on X.
He desired to deepen his focus on AI alignment and engage in more hands-on technical work. Schulman believes that Anthropic will offer new perspectives and opportunities to pursue his research interests alongside like-minded experts.
In addition to Schulman’s departure, OpenAI has confirmed that president and co-founder Greg Brockman is taking an extended leave until the end of the year. Brockman, who has been with the company for nine years, plans to use this time to relax and recharge.
This news follows the earlier exit of Peter Deng, a product manager who joined OpenAI last year after leading product teams at Meta, Uber, and Airtable. The Information had already reported Deng’s departure.
A spokesperson for OpenAI expressed gratitude for Schulman’s contributions, highlighting his role in advancing alignment research and establishing a foundation for future innovations. Despite Schulman’s departure, the spokesperson conveyed confidence that OpenAI’s alignment research efforts would continue to thrive.
Schulman’s involvement with OpenAI began after he completed his Ph.D. in electrical engineering and computer sciences at UC Berkeley.
His work at OpenAI was instrumental in creating ChatGPT, where he led the reinforcement training organization. Following the departure of Jan Leike, another AI safety researcher who also joined Anthropic, Schulman took over as head of OpenAI’s alignment science efforts and was a member of the newly formed safety committee.
Despite ongoing controversies surrounding OpenAI’s approach to AI safety research, Schulman emphasized that his decision to leave was not due to a lack of support from the company. Instead, he framed it as a personal choice driven by his desire to focus his efforts on new research challenges at Anthropic.
With Schulman and Brockman’s exit, only three people from OpenAI’s original founding team remain: Sam Altman, who is still the CEO, and Wojciech Zaremba, who leads language and code generation. Because he is on probation, Brockman is technically still a part of the organization.
Altman expressed his gratitude and admiration for Schulman’s contributions, describing him as a brilliant researcher, a deep thinker about products and society, and a great friend to the team.
As OpenAI continues to navigate these significant changes, its leadership and remaining founders face the challenge of maintaining momentum and ensuring the continuity of their ambitious AI research and development efforts.
