OpenAI’s dedicated superalignment team, created to study how future artificial intelligence systems might be steered or controlled, has been dissolved. The company confirmed that the group’s work will be absorbed into other OpenAI research efforts.
The move lands at a sensitive moment for the company. It follows the exit of Ilya Sutskever, OpenAI’s chief scientist and one of its cofounders, the resignation of Jan Leike, and several other departures tied to AI safety, policy, and governance work.
A team built for the hardest version of AI safety
OpenAI announced the superalignment team in July last year. Its job was to prepare for the possibility of supersmart artificial intelligence capable of outwitting and overpowering its creators.
Sutskever was named as one of the team’s coleads. OpenAI also said the team would receive 20 percent of its computing power, making the group one of the company’s most visible commitments to long-term AI risk research.
The team was not the only part of OpenAI focused on keeping AI under control. But it was publicly positioned as the main group working on the most distant and difficult version of that problem: how to handle systems that could become more capable than humans.
OpenAI’s own announcement for the team stated: “Currently, we don't have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue.” That framing made the group a symbol of the company’s stated concern that advanced AI could create risks well beyond today’s chatbot behavior.
The departures changed the picture
The team’s dissolution came after several researchers connected to it left or were reported to have left. Sutskever’s departure drew major attention because of his central role in OpenAI’s history. He helped CEO Sam Altman start OpenAI in 2015 and helped set the research direction that led to ChatGPT.
Sutskever was also one of the four board members who fired Altman in November. Altman returned as CEO five chaotic days later after a mass revolt by OpenAI staff and a deal in which Sutskever and two other company directors left the board.
Hours after Sutskever’s departure was announced on Tuesday, Jan Leike, the former DeepMind researcher who colead the superalignment team, posted on X that he had resigned. Neither Sutskever nor Leike responded to requests for comment in the source article.
Sutskever did not explain his decision to leave, but he voiced support for OpenAI’s direction in a post on X. He wrote: “The company’s trajectory has been nothing short of miraculous, and I’m confident that OpenAI will build AGI that is both safe and beneficial” under its current leadership.
Leike offered a different signal in a thread on X on Friday. He said his resignation followed a disagreement over OpenAI’s priorities and the resources allocated to his team.
“I have been disagreeing with OpenAI leadership about the company's core priorities for quite some time, until we finally reached a breaking point,” Leike wrote. “Over the past few months my team has been sailing against the wind. Sometimes we were struggling for compute and it was getting harder and harder to get this crucial research done.”
AI governance work also appears to be shifting
The breakup of the superalignment team adds to signs of a broader shakeout inside OpenAI after last November’s governance crisis. Two researchers on the team, Leopold Aschenbrenner and Pavel Izmailov, were dismissed for leaking company secrets, The Information reported last month.
Another member of the team, William Saunders, left OpenAI in February, according to an internet forum post in his name. Two more OpenAI researchers working on AI policy and governance also appear to have left recently.
Cullen O'Keefe left his role as research lead on policy frontiers in April, according to LinkedIn. Daniel Kokotajlo, who has coauthored several papers on the dangers of more capable AI models, “quit OpenAI due to losing confidence that it would behave responsibly around the time of AGI,” according to a posting on an internet forum in his name.
None of the researchers who apparently left responded to requests for comment in the source article. OpenAI declined to comment on the departures of Sutskever or other members of the superalignment team, or on the future of its long-term AI risk work.
The company’s research on risks tied to more powerful models will now be led by John Schulman. He coleads the team responsible for fine-tuning AI models after training.
Why the timing is drawing attention
OpenAI’s charter binds it to safely developing artificial general intelligence, described as technology that rivals or exceeds humans, and doing so for the benefit of humanity. Sutskever and other leaders have often spoken about the need to proceed cautiously.
At the same time, OpenAI has been early to develop and publicly release experimental AI projects. That tension is part of why the fate of the superalignment team matters: it was a dedicated structure for studying long-range risks while the company continued to ship increasingly capable systems.
OpenAI became the most prominent and closely watched technology company on the planet after ChatGPT. As researchers and policymakers considered ChatGPT and the prospect of more capable AI, concern about AI harming humans or humanity as a whole became less controversial.
The source article notes that existential anxiety around AI has cooled, and AI has yet to make another massive leap. But AI regulation remains a hot topic, and OpenAI has continued to introduce systems that could alter how people relate to the technology.
GPT-4o keeps the safety debate alive
The departures of Sutskever and Leike came shortly after OpenAI’s latest big reveal: a new “multimodal” AI model called GPT-4o. The model allows ChatGPT to see the world and converse in a more natural and humanlike way.
A livestreamed demonstration showed the new version of ChatGPT mimicking human emotions and even attempting to flirt with users. OpenAI has said it will make the new interface available to paid users within a couple of weeks.
There is no indication that the recent departures have anything to do with OpenAI’s work on more humanlike AI or product releases. Still, the latest advances raise ethical questions around privacy, emotional manipulation, and cybersecurity risks.
OpenAI maintains another research group called the Preparedness team, which focuses on those issues. For now, the company’s long-term AI risk work is no longer housed in the standalone superalignment team that was announced with significant public attention in July last year.