Anthropic Launches The Anthropic Institute to Address AI's Societal Challenges

Anthropic Institute formally launched

Announced March 12, 2026, the Institute is a new organizational unit dedicated to researching and communicating the societal consequences of increasingly powerful AI systems, intended to serve both researchers and the general public.

Jack Clark named Head of Public Benefit to lead the Institute

Clark is an Anthropic co-founder, giving the Institute significant internal standing. His new title signals Anthropic is elevating public-benefit work to a named executive function rather than treating it as a side effort.

Three existing research teams consolidated under the Institute

The Frontier Red Team (stress-testing AI capabilities), Societal Impacts (real-world AI use studies), and Economic Research (jobs and economy impact tracking) are merged under a single umbrella, improving coordination across previously siloed efforts.

New teams incubated for AI progress forecasting and legal system interactions

The Institute is standing up groups focused on predicting AI progress trajectories and studying how AI intersects with legal frameworks — areas not previously formalized at Anthropic.

Matt Botvinick hired to lead AI and rule-of-law research

Botvinick is a Yale Law School Resident Fellow, former Google DeepMind Senior Research Director, and former Princeton professor in Neural Computation — a cross-disciplinary profile spanning AI research and legal academia.

Anton Korinek leads economic restructuring research

Korinek is a UVA Economics professor on leave, focused on how AI-driven economic shifts affect workers and industries. His inclusion underscores the Institute's emphasis on labor displacement as a near-term concern.

Zoë Hitzig joins from OpenAI to study social and economic impacts

Hitzig brings direct competitor experience from OpenAI and focuses on how AI reshapes social and economic structures, adding policy-facing social science depth to the Institute's lineup.

Anthropic forecasts far more dramatic AI progress within two years

Anthropic states it expects far more dramatic progress in the next two years than in the previous five, and raises questions about who should govern recursive self-improvement if it begins to occur — implying the window for adaptive frameworks is short.

Institute commits to candid public reporting and direct engagement with displaced workers

This positions Anthropic differently from typical corporate research labs by pledging transparency about frontier findings and engaging directly with communities facing economic displacement rather than only publishing academic outputs.