Ex-OpenAI Safety Leader Leike To Join Rival Anthropic

0 345

Jan Leike, the former co-head of an OpenAI safety team focused on long-term risks, is joining its artificial intelligence rival Anthropic, Leike said in a post on X Tuesday.

Leike left the company earlier this month after Ilya Sutskever, the other leader of OpenAI’s Superalignment team, resigned from the company. The team they once lead was subsequently dissolved.

As he departed Leike criticized OpenAI, saying that “safety culture and processes have taken a backseat to shiny products.” On Tuesday, OpenAI formed a new safety team featuring Chief Executive Officer Sam Altman.

Leike’s work at Anthropic will focus on similar topics to what he worked on at OpenAI, he said — safety issues related to the control of so-called superhuman AI models. Such AI models do not yet exist, but companies such as OpenAI and Anthropic are conducting research into how they could be controlled if they are built in the future.

Anthropic, formed by former OpenAI staffers, has billed itself as a more safety-focused AI company. Leike did not immediately respond to a request for comment.