Anthropic launched a new internal think tank today called the Anthropic Institute, combining three existing research teams into a single unit focused on understanding AI's large-scale societal implications.

The institute merges Anthropic's societal impacts team, frontier red team, and economic research team. Its stated research agenda covers questions like what happens to jobs and economies as AI advances, whether AI makes society safer or more dangerous, and whether humans can retain meaningful control over increasingly powerful systems.

Co-founder Jack Clark is moving to lead the institute as head of public benefit, a new title, after more than five years as head of public policy. Sarah Heck, formerly head of external affairs, takes over the policy function. Anthropic is also opening a Washington, DC office as part of the restructuring.

The institute launches with roughly 30 founding members, including Matt Botvinick, formerly of Google DeepMind; economist Anton Korinek of the University of Virginia; and Zoe Hitzig, who left OpenAI following its decision to introduce ads into ChatGPT.

The announcement comes amid a tense standoff between Anthropic and the US government. The Trump administration designated Anthropic as a military supply-chain risk after the company refused to allow its Claude models to be used for mass domestic surveillance and fully autonomous lethal weapons systems. Anthropic subsequently sued the Department of Defense, alleging the blacklist was unconstitutional retaliation for protected speech under the First and Fifth Amendments. A preliminary injunction hearing took place today before Judge Rita Lin in a California district court.

Clark said the institute's launch had been in development for months and that the recent conflict with the Pentagon "has affirmed" Anthropic's decision to release more public information about AI's societal implications. Anthropic is also reportedly planning an IPO in 2026.