Former OpenAI Safety Lead Joins Anthropic to Head New AI Superalignment Team

Jan Leike Takes on New Role at Anthropic

Anthropic, an emerging player in AI research, has hired Jan Leike, former AI safety lead at OpenAI, to spearhead a new “superalignment” team. Leike’s move follows his recent resignation from OpenAI, where he publicly criticized the company’s approach to AI safety.

Focus on AI Safety and Security

In a recent post on X, Leike outlined the primary objectives of his new role at Anthropic. His team will concentrate on critical areas of AI safety, including scalable oversight, weak-to-strong generalization, and automated alignment research. These efforts aim to enhance the predictability and desirability of large-scale AI behavior.

Reporting Structure and Team Realignment

According to a source familiar with the development, Leike will report directly to Jared Kaplan, Anthropic’s chief science officer. Researchers at Anthropic currently focused on scalable oversight will transition to report to Leike as his team begins its operations.

Parallels with OpenAI’s Superalignment Team

Leike’s mission at Anthropic echoes his previous work with OpenAI’s Superalignment team, which was dissolved earlier this year. The Superalignment team had aimed to tackle the core technical challenges of controlling superintelligent AI within four years but faced internal challenges and leadership constraints.

Anthropic’s Commitment to AI Safety

Anthropic has consistently positioned itself as a more safety-oriented alternative to OpenAI. The company’s CEO, Dario Amodei, formerly served as VP of research at OpenAI. Amodei left OpenAI over disagreements regarding the company’s increasing commercial focus. He founded Anthropic with several ex-OpenAI colleagues, including former policy lead Jack Clark.

Jan Leike’s transition to Anthropic marks a significant step in the ongoing evolution of AI safety research. By leading a dedicated superalignment team, Leike aims to address some of the most pressing challenges in AI safety, reinforcing Anthropic’s commitment to developing secure and predictable AI technologies.

DIGITAL MARKETING: New tips, resources to grow your business

REVIEW: Apple iPad (10th Generation) Review: The Best Tablet Under $300