John Schulman, a co-founder of OpenAI, announced his departure from the Microsoft-backed company to join Anthropic, an artificial intelligence startup funded by Amazon. Schulman’s decision to leave OpenAI, where he has been a key figure for nearly nine years, reflects his desire to deepen his focus on AI alignment and return to hands-on technical work.
Schulman emphasized that his departure was not due to a lack of support for alignment research at OpenAI, but rather a personal choice to pursue new perspectives and research opportunities at Anthropic. He joined OpenAI shortly after completing his Ph.D. in computer science from the University of California, Berkeley in 2016 and has been integral to the company’s development and success.
During his tenure at OpenAI, Schulman co-led the post-training team responsible for refining AI models for the ChatGPT chatbot and the programming interface for third-party developers. In June, he was appointed head of alignment science and joined a safety and security committee to provide advice to the board. His contributions were significant in shaping the early strategy of OpenAI.
Schulman’s departure comes in the wake of recent structural changes at OpenAI, including the disbanding of the superalignment team. This team, focused on ensuring human control over highly capable AI systems, saw its leaders Jan Leike and Ilya Sutskever leave the company earlier this year. Leike also joined Anthropic, while Sutskever is involved in founding Safe Superintelligence Inc.
Anthropic, established by former OpenAI staff members in 2021, is emerging as a significant player in the competitive landscape of generative AI, alongside giants like Amazon, Google, and Meta. Schulman expressed confidence in OpenAI’s future success, praising the talent and leadership within the organization. He highlighted the company’s commitment to safety and alignment research, stating that OpenAI leaders have been very committed to investing in this area.
In related developments, Greg Brockman, another co-founder of OpenAI and its president, recently announced a sabbatical for the remainder of the year. OpenAI continues to collaborate with the US AI Safety Institute to advance the science of AI evaluations and maintains a commitment to allocating 20% of its computing resources to safety initiatives.
