Anthropic hires former OpenAI safety lead to head up new team
Jan Leike, a leading AI researcher who earlier this month resigned from OpenAI before publicly criticizing the companyâs approach to AI safety, has joined OpenAI rival Anthropic to lead a new âsuperalignmentâ team.
In a post on X, Leike said that his team at Anthropic will focus on various aspects of AI safety and security, specifically âscalable oversight,â âweak-to-strong generalizationâ and automated alignment research.
A source familiar with the matter tells TechCrunch that Leike will report directly to Jared Kaplan, Anthropicâs chief science officer, and that Anthropic researchers currently working on scalable oversight â techniques to control large-scale AIâs behavior in predictable and desirable ways â will move to report to Leike as Leikeâs team spins up.
In many ways, Leikeâs team sounds similar in mission to OpenAIâs recently-dissolved Superalignment team. The Superalignment team, which Leike co-led, had the ambitious goal of solving the core technical challenges of controlling superintelligent AI in the next four years, but often found itself hamstrung by OpenAIâs leadership.
Anthropic has often attempted to position itself as more safety-focused than OpenAI.
Anthropicâs CEO, Dario Amodei, was once the VP of research at OpenAI, and reportedly split with OpenAI after a disagreement over the companyâs direction â namely OpenAIâs growing commercial focus. Amodei brought with him a number of ex-OpenAI employees to launch Anthropic, including OpenAIâs former policy lead Jack Clark.