OpenAI co-founder John Schulman announced via X that he would leave the Microsoft-backed AI firm to join Anthropic, a competing artificial intelligence startup funded by Amazon.
This comes less than three months after OpenAI disbanded its superalignment team. The division was dedicated to ensuring that AI systems remain controllable despite surpassing human capabilities in various tasks.
Reasons For The Transition
Schulman, who has been instrumental in refining AI models for OpenAI’s ChatGPT and its programming interface for third-party developers, was previously the head of alignment science. In June, OpenAI said Schulman would transition to a safety and security committee, providing guidance to the board. Schulman, who earned his Ph.D. in computer science from the University of California, Berkeley in 2016, has been with OpenAI since then.
Schulman wrote on X:
“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.”
He clarified that his departure was not due to a lack of support for AI safety initiatives at OpenAI.
He said:
“On the contrary, company leaders have been very committed to investing in this area.”
Impact Of Recent Departures
This announcement follows recent departures from OpenAI, including Jan Leike, the leader of the superalignment team, and co-founder Ilya Sutskever. Leike has also joined Anthropic, while Sutskever is involved in launching a new venture, Safe Superintelligence Inc.
Since its establishment in 2021, Anthropic has constantly competed with OpenAI to develop the most advanced generative AI models capable of producing human-like text. Other major tech players, such as Amazon, Google, and Meta, are also advancing.
Leike expressed enthusiasm about working with Schulman again, highlighting the collaboration as a positive development. OpenAI’s CEO Sam Altman acknowledged that Schulman’s insights were instrumental in shaping the company's early strategies.
Schulman’s departure and the broader changes at OpenAI follow a period of upheaval last November when the board ousted Altman. This decision led to employee protests and the resignation of board members, including Sutskever, Tasha McCauley, and Helen Toner. Altman was eventually reinstated, and OpenAI expanded its board.
Toner had previously criticized Altman for allegedly providing inaccurate information about the company’s safety processes. An independent review by the law firm WilmerHale revealed that the board had not prioritized product safety in its decision-making.
Future Developments
Last week, Altman announced that OpenAI is collaborating with the US AI Safety Institute to provide early access to its upcoming foundation model, aiming to advance the science of AI evaluation. OpenAI remains committed to dedicating 20 percent of its computing resources to safety initiatives.
On the same day as Schulman’s announcement, Greg Brockman, OpenAI’s president and co-founder, disclosed that he would be taking a sabbatical for the remainder of the year.
