Anthropic sues after Pentagon labels AI firm a “supply chain risk”

Anthropic has filed a lawsuit in a California federal court challenging the Defense Department’s decision to label the artificial intelligence company a "supply chain risk," a designation that could effectively bar it from future military contracts. The Pentagon notified Anthropic via a letter dated March 4 and said the designation was effective immediately, escalating a weekslong standoff over the terms under which the company’s systems may be used by the U.S. government.

Anthropic’s chief executive, Dario Amodei, confirmed the company received the Pentagon letter and said, "We do not believe this action is legally sound." The company said it saw no choice but to contest the designation in court. An internal post from the company later apologized for an earlier strongly worded statement, saying, "It was a difficult day for the company, and I apologize for the tone of the post. It does not reflect my careful or considered views. It was also written six days ago, and is an out-of-date assessment of the current situation."

The dispute grew out of last year’s Pentagon pilot program in which Anthropic, OpenAI, Google and xAI tested A.I. for defense use. Anthropic was the only firm that deployed its systems on classified systems and became widely used by defense officials. The company sought explicit contractual guardrails to bar its technology from being used for domestic mass surveillance or deployed in autonomous weapons without human involvement. The Defense Department pushed back, insisting that contractors could not unilaterally limit military use and directing that contracts adopt "any lawful use" language.

Policy friction intensified after a January memorandum directed that DoD agreements include that "any lawful use" clause. Department of Defense leadership, including Secretary of Defense Peter Hegseth, met with Amodei at the Pentagon on Feb. 24 in an effort to resolve the dispute; the meeting lasted less than an hour and was described by people familiar with it as terse. At times the department threatened to invoke the Defense Production Act to compel cooperation, though officials later pursued the supply chain risk designation.

The supply chain label has historically been applied to foreign companies the United States views as national security threats; it has not previously been used against an American firm. That precedent raises immediate legal questions. Anthropic intends to seek judicial review, and its challenge is likely to invoke the Administrative Procedure Act. Legal analysts note a 2021 ruling in Luokung Technology v. Department of Defense, where a court vacated a DoD designation for lack of basis or procedural fairness, as a relevant precedent that could open the department’s decision to close scrutiny.

Operationally, the designation targets government contracting: it could bar Anthropic from new DoD business and trigger removal from systems that currently rely on its models. The scope is narrower for non-government customers: the company’s Claude chatbot and other products remain available to the public and to private-sector partners. Microsoft, after legal review, said it will continue to use Anthropic’s technology for non-defense projects.

Intelligence agencies that use Anthropic’s systems, including the CIA, privately urged both sides to negotiate, according to officials familiar with the matter, highlighting the immediate national security stakes. The court filing and the DoD’s underlying letter will determine how quickly the dispute moves from negotiation to a legal test of Pentagon authority over domestic technology providers. The outcome will shape how Silicon Valley and Washington resolve guardrails, control and access for the next generation of military A.I. providers.