Anthropic launches the Anthropic Institute for public benefit research

Anthropic today announced The Anthropic Institute, a new internal organization placing its Frontier Red Team, Societal Impacts, and Economic Research teams under a single structure led by co-founder Jack Clark, who takes the new title of Head of Public Benefit. The Institute's mandate is to produce research on AI safety, economic effects, and societal risk directed at policymakers and the public -- not to contribute directly to model development. Three external researchers join as founding members: Matt Botvinick (Yale Law, ex-Google DeepMind, AI and rule of law), Anton Korinek (UVA economics, economic impacts of transformative AI), and Zoe Hitzig (ex-OpenAI, bridging economic research and model development practice).

The announcement arrives at a loaded moment for Anthropic. In February 2026, the company revised its Responsible Scaling Policy (RSP v3.0) and dropped a prior pledge to halt training if certain safety thresholds were crossed -- a change criticized by some safety researchers. Separately, the Pentagon designated Anthropic a "supply chain risk" after the company declined to support weapons and surveillance applications. The Anthropic Institute consolidates existing work rather than creating new spending lines; Anthropic has not disclosed a dedicated budget, though the company closed a $30 billion funding round in early 2026 and reports approximately $20 billion in annualized revenue. The Institute is organizationally distinct from Anthropic's technical safety teams (alignment and interpretability research) and from its separate Public Policy organization, which operates under Sarah Heck out of a new Washington, DC office.