More Churn in OpenAI’s Top Leadership

Apparently there’s more security in the top jobs in Westeros than there is at OpenAI. While the turnover is less violent in the San Francisco offices of the artificial intelligence market leader than the bloody castles of the HBO fantasy epic, both seem none too stable. OpenAI co-founder John Schulman reportedly left the company, and another co-founder, Greg Brockman, is taking a year off to “relax and recharge.”

This is tough news for any startup, but Schulman’s departure stings more since he’s leaving to join fierce rival Anthropic. That means three of the 11 co-founders of OpenAI have left within the last year. And don’t forget the antics of co-founder and former board member Elon Musk, who just filed a new lawsuit against the company on alleged “racketeering” grounds.  

Tech news site The Information reported the departures, and also mentioned that Peter Deng, a product manager who previously had senior jobs at Meta, Uber, and Airtable, and only joined OpenAI last year, has also left. The site did point out that the leave of absence and the exiting staff “appear to be unrelated to each other,” possibly because they happened at different times. TechCrunch notes the company has now confirmed that Deng left “some time ago.”

Though seemingly unrelated, three senior management departures in a short window is a concern for any business, and shows that the “company’s leadership has yet to stabilize,” The Information says.

Schulman’s departure may be the most painful. His website says he was co-lead of the “post-training team” at the company, where OpenAI works to “fine-tune the models that get deployed in ChatGPT and the OpenAI API.” He posted about his move on X, TechCrunch reports, highlighting he wanted more hands-on work on AI “alignment,” which is making sure as AIs get smarter, they do so for the benefit of humans rather than causing harm.

In his long X post, Schulman says he’s moving to Anthropic because he thinks he “can gain new perspectives and do research alongside people deeply engaged with the topics” he’s interested in. He also noted his move has nothing to do with safety issues at OpenAI: “On the contrary, company leaders have been very committed to investing in this area,” he said, adding “my decision is a personal one.”

Nevertheless, Schulman’s exit may reignite old worries that the company is pursuing profits at the expense of AI safety. This played a role in the OpenAI board’s short-lived ouster of CEO and co-founder Sam Altman last autumn. The complex spat also saw the company’s “superalignment” safety team leader and co-founder Ilya Sutskever leave the office, never to return. Sutskever finally quit OpenAI in May to launch his own safety-centric AI company. His departure went public mere hours before safety team co-leader Jan Leike announced he too was leaving.

OpenAI has since taken pains to point out that it is still very conscious of the risks AI represents, and has made public demonstrations that it’s working on safety, including making changes to its board. TechCrunch reports that Altman has also recommitted to an earlier company pledge to have 20 percent of its computer power dedicated to working on safety issues. However, employees recently called for whistleblower protections to keep the company honest in the face of potential future safety issues.

What can you and your company learn from all this? One lesson seems clear: keeping things simple, straightforward, and scandal-free in the executive suite, with leaders all aligned to company goals, means you may avoid complex, volatile leadership issues arising later.