OpenAI Co-Founder Schulman Joins Anthropic; Brockman Begins Extended Leave of Absence

John Schulman Leaves OpenAI for Anthropic; Greg Brockman Takes Extended Leave

John Schulman, a co-founder of OpenAI, has transitioned to the rival AI startup Anthropic. Additionally, OpenAI President and co-founder Greg Brockman is taking a much-needed sabbatical until the year's end, stating a desire to "relax and recharge" after nearly a decade with the company.

In a post on X, Brockman shared, “I’m taking a sabbatical through the end of the year. It’s my first chance to relax since co-founding OpenAI. Although our mission is still in progress, we have more work to do in building safe AGI.”

Peter Deng, a product manager who joined the ranks of OpenAI in the past year after leading projects at Meta, Uber, and Airtable, also left the company some time ago. The departures of Brockman and Deng were initially reported by The Information.

A spokesperson expressed gratitude for Schulman's contributions: “We appreciate John’s efforts as a founding team member and his dedication to advancing alignment research. His passion and hard work have laid a strong groundwork for future innovations both at OpenAI and throughout the AI industry.”

In today’s announcement on X, Schulman explained that his departure stems from a desire to delve deeper into AI alignment—the discipline focused on ensuring AI operates as intended. He aims to engage in more hands-on technical work at Anthropic, alongside experts in his field of interest. “I’ve decided to pursue this goal at Anthropic, where I believe I can gain new insights and collaborate with others passionate about these topics," Schulman stated. “I am certain that OpenAI and the talented teams I was part of will continue to excel in my absence.”

Schulman's journey with OpenAI began after earning a PhD in electrical engineering and computer sciences from UC Berkeley. He played a crucial role in developing the AI chatbot platform ChatGPT, leading OpenAI’s reinforcement training organization, which fine-tunes generative AI models to align with human instructions.

Following the exit of AI safety researcher Jan Leike, who has also joined Anthropic, Schulman assumed leadership of OpenAI’s alignment science efforts, known as the “post-training” team. He was also involved in the recently established safety committee, though it remains uncertain who will take over his position.

Despite existing controversies surrounding OpenAI, particularly regarding its stance on AI safety research, Schulman emphasized that his departure was not due to a lack of support. “Company leaders have been profoundly committed to investing in alignment research,” he stated. “My decision to leave is personal, motivated by how I wish to direct my focus in the next chapter of my career.”

With Schulman's exit, only three of OpenAI's original eleven founders are still with the organization: CEO Sam Altman, Greg Brockman, and Wojciech Zaremba, who leads language and code generation.

Altman expressed his gratitude on X, stating, “Thank you for everything you’ve done for OpenAI! You are a brilliant researcher and a deep thinker about product and society. Most importantly, you are a great friend to all of us. We will miss you tremendously and strive to make you proud of this organization.”

Update: This story was initially published at 5:38 p.m. and has since been updated to confirm the departures of Brockman and Deng.