Key Takeaways
OpenAI co-founder and leading AI alignment researcher John Schulman has left his role at Anthropic just six months after joining the startup.
Schulman’s departure was confirmed by Anthropic on Wednesday, Feb. 5, raising the possibility that he may follow other OpenAI co-founders and form a new AI venture.
One of OpenAI’s eleven co-founders, Schulman, led the reinforcement learning team that developed ChatGPT.
Schulman’s great contribution to the project was applying reinforcement learning with human feedback (RLHF) to OpenAI’s language models.
As he explained in 2023: “The idea was to align our models with human preference — try to get them to actually listen to us and try to do what we want.”
In the years since then, Schulman has continued to focus on the notion of alignment.
In August 2024, he left OpenAI amid a wave of departures that included fellow co-founders Jan Leike and Mira Muratti.
At the time, he said he was taking up the role at Anthropic to “focus on AI alignment” and “return to hands on technical work.”
Although Schulman has not yet disclosed what he plans to do next, Anthropic’s chief science officer Jared Kaplan said : “We are sad to see John go but fully support his decision to pursue new opportunities.”
One possibility is that he could follow several other OpenAI co-founders who have gone on to form their own AI startups.
These include Ilya Sutskever, who founded Safe Superintelligence, Andrej Karpathy, who launched Eureka Labs, and Vicki Cheung, who helped found Gantry.
While Schulman’s tenure at Anthropic was especially short lived, hopping between AI labs is commonplace among the field’s leading researchers.
OpenAI itself initially poached top talent from Google DeepMind.
Meanwhile, Schulman isn’t the only former OpenAI employee to find a home at Anthropic. Others include fellow OpenAI co-founders Jared Kaplan and Durk Kingman, the leading AI safety researcher Jan Leike, and Anthropics CEO and President Dario and Daniela Amodei.