OpenAI Disbands AI Safety Team After Key Leaders Depart
OpenAI dissolves its Superalignment team, which focused on controlling advanced AI, just one year after its creation. The breakup follows the departure of team leaders Ilya Sutskever and Jan Leike, who criticizes the company for prioritizing products over safety.
OpenAI disbands its Superalignment team, a group dedicated to managing the long-term risks of artificial intelligence, just one year after its highly publicized launch. The company confirms that members of this specialized team are reassigned to various other divisions across the organization. This structural shift occurs only days after the team's co-leaders, co-founder Ilya Sutskever and researcher Jan Leike, officially announce their exits from the Microsoft-backed startup.
Jan Leike publicly criticizes OpenAI's current direction, stating that safety culture and processes take a backseat to shiny products. He explains that he disagrees with OpenAI leadership about core priorities, noting that his team struggles to secure adequate computing resources to conduct crucial research. Leike argues that the company needs to focus significantly more of its bandwidth on security, monitoring, preparedness, safety, and societal impact.
OpenAI does not provide a direct comment on the team's dissolution but points to a recent social media post from CEO Sam Altman, who acknowledges that the company has more work to do regarding safety. Co-founder Greg Brockman releases a joint statement with Altman emphasizing OpenAI's efforts to raise awareness about the risks and opportunities of artificial general intelligence. The breakup of the Superalignment group marks a significant shift for a company that originally promises to dedicate twenty percent of its computing power to this initiative over four years.