Anthropic sues over 'Claude' use
Anthropic is in a San Francisco legal battle with the U.S. Department of Defense over the DoD's use of its 'Claude' model for military targeting—accusing the Pentagon of mass-approving AI-generated strike lists without adequate human review. The dispute echoes revelations about the Pentagon's Project Maven push to embed AI in combat and has intensified warnings that tech firms risk losing control of high-stakes generative systems. (techpolicy.press) (npr.org)
The Defense Department sent formal letters notifying Anthropic of a “supply‑chain risk” designation dated March 3, 2026, and the agency publicly formalized the designation on March 5, 2026. (mayerbrown.com)) Anthropic filed two suits on March 9, 2026 — a 48‑page complaint in the U.S. District Court for the Northern District of California and a separate petition in the U.S. Court of Appeals for the D.C. Circuit — challenging the legality of the designation and asserting constitutional claims. (lawfaremedia.org)) The federal case in San Francisco is assigned to U.S. District Judge Rita F. Lin, who at a March 24 hearing said the blacklisting “looks like an attempt to cripple Anthropic” and pressed government counsel on the legal basis for the label. (courtlistener.com)) The designation forces compliance steps that could require major defense contractors — including Amazon, Microsoft and Palantir — to certify they are not using Anthropic’s systems in Pentagon work, and agencies have roughly 180 days to phase the company’s products out. (cnbc.com)) Anthropic’s court filings warn the blacklist jeopardizes billions in revenue and say federal and private contracts have already been canceled or put at risk, framing the case as immediate commercial harm. (cnbc.com)) Microsoft, Google, Amazon, Apple, and other tech firms have filed amicus briefs or public statements supporting Anthropic, while more than 30 employees from OpenAI and Google DeepMind and retired military officials have submitted supportive filings. (techcrunch.com)) Anthropic’s own announcement shows a July 14, 2025 prototype agreement with the Defense Department with a $200 million ceiling for AI work, and U.S. reporting says the company’s model was used by military personnel in operations in Iran in early March 2026. (anthropic.com))
