Jan Leike, a prominent AI researcher who recently left OpenAI due to concerns over the company’s approach to AI safety, has joined Anthropic, a competitor of OpenAI, to head a new “superalignment” team. This underscores a significant shift within the AI safety community and highlights Anthropic’s commitment to prioritizing safety in AI development.
Jan Leike’s New Role at Anthropic
At Anthropic, Leike will lead efforts focusing on “scalable oversight,” “weak-to-strong generalization,” and automated alignment research. These areas are critical for ensuring that AI systems operate safely and are aligned with human values, especially as AI technologies become increasingly capable and autonomous. Leike’s role will involve direct reporting to Jared Kaplan, Anthropic’s Chief Science Officer, indicating the strategic importance of the superalignment team within the company.
Jan Leike’s departure from OpenAI was notable, particularly because he voiced concerns publicly about the organization’s trajectory in AI safety. His move to Anthropic is seen as a natural transition given his commitment to AI safety, aligning with Anthropic’s mission to develop safer AI technologies.
I'm excited to join @AnthropicAI to continue the superalignment mission!
— Jan Leike (@janleike) May 28, 2024
My new team will work on scalable oversight, weak-to-strong generalization, and automated alignment research.
If you're interested in joining, my dms are open.
Leike’s new position at Anthropic, and the focus of his team, mirrors the objectives of the now-dissolved Superalignment team at OpenAI, which he co-led. The dissolution of this team at OpenAI has been cited as part of broader tensions within the company regarding its prioritization of commercial objectives over safety concerns.
Anthropic’s Safety-First Philosophy
Under the leadership of Dario Amodei, Anthropic has been vocal about its safety-centric approach to AI development. Amodei, a former VP of research at OpenAI, founded Anthropic after reportedly splitting from OpenAI due to disagreements over the company’s direction, particularly its increasing focus on commercialization.
Jan Leike’s move to Anthropic and the formation of a new superalignment team signal a potential shift in the landscape of AI development, particularly in how safety concerns are integrated into the research and deployment of AI technologies.
Jan Leike’s appointment at Anthropic represents a significant development in the AI community, highlighting ongoing debates and shifts regarding AI safety and governance. As AI continues to evolve, the approaches taken by companies like Anthropic and OpenAI will critically influence not only the advancement of technology but also the broader societal implications of AI.