OpenAI co-founder John Schulman on Monday announced his departure from the Microsoft-backed firm to join the AI startup Anthropic, which is supported by Amazon.
"This choice stems from my desire to deepen my focus on AI alignment, and to start a new chapter of my career where I can return to hands-on technical work," Schulman said in a post on X (formerly Twitter).
His departure follows OpenAI's recent dissolution of its superalignment team, a group dedicated to ensuring human control over advanced AI systems, according to Reuters.
OpenAI's co-founder and president Greg Brockman also announced on Monday that he would take a sabbatical for the remainder of the year.
At OpenAI, Schulman played a key role in the post-training team that fine-tuned AI models for the ChatGPT chatbot and developed a programming interface for third-party developers.
According to his website, Schulman, who earned his Ph.D. in computer science from UC Berkeley in 2016, was set to join a safety and security committee providing guidance to OpenAI's board as head of alignment science.
Adding that his departure was not due to a lack of support for AI alignment efforts at OpenAI, Schulman praised the company leaders for their commitment to the project.
OpenAI's CEO Sam Altman, in a post, acknowledged Schulman's impact on the company's early strategies.
Schulman's departure comes in the wake of significant upheaval at OpenAI, including Altman's brief ousting last November.
The decision, which led to employee protests and the resignation of board members Ilya Sutskever, Tasha McCauley, and Helen Toner, was reversed with Altman's reinstatement and the expansion of the board.
In a podcast, Toner accused Altman of providing the board with inaccurate information regarding the company's safety protocols.
Later, an independent review by law firm WilmerHale concluded that the board was not primarily concerned with product safety when it decided to remove Altman.
Last week, Altman announced that OpenAI is collaborating with the US AI Safety Institute to provide early access to its next foundational model. He also reaffirmed OpenAI's pledge to dedicate 20% of its computing resources to safety initiatives.
Schulman's departure also comes after Jan Leike and Sutskever, who were part of the superalignment team, exited OpenAI this year. While Leike joined Anthropic, Sutskever launched his own AI firm Safe Superintelligence Inc.
Founded in 2021 by former OpenAI staff, Anthropic has quickly become a competitor in the race to develop top-tier generative AI models, alongside tech giants like Amazon, Google, and Meta.