Anthropic on Monday asked a federal court in California to invalidate a Pentagon decision that placed the artificial intelligence company on a national security supply-chain risk list, escalating a high-profile clash over restrictions on military uses of its technology.
The company said the designation, announced last week by Defense Secretary Pete Hegseth, was unlawful and violated its rights to free speech and due process. In its filing, Anthropic requested that a judge rescind the designation and bar federal agencies from enforcing the supply-chain restriction.
Anthropic argued in court papers that the move by the Department of Defense is unprecedented and an improper exercise of governmental power. "These actions are unprecedented and unlawful. The Constitution does not allow the government to wield its enormous power to punish a company for its protected speech," the company said in the filing.
The Pentagon formalized the supply-chain risk label after months of discussions with Anthropic over the company's policies limiting certain uses of its models. According to the Pentagon, the designation responds to Anthropic's refusal to remove protective guardrails that the company says are intended to prevent its technology from being used for fully autonomous weapon systems or domestic surveillance of Americans.
Officials contend that U.S. law, not the policies of a private company, should determine how to defend the country, and they have stressed the need for full flexibility to employ AI for "any lawful use." The Defense Department has argued that limitations placed by Anthropic could hinder military operations and, in their view, potentially endanger American lives.
Anthropic has maintained a contrary position. The company told the court that even the best current AI models are not reliable enough for deployment as fully autonomous weapons and that such applications would be dangerous. Anthropic also drew a firm line on domestic surveillance, asserting that using its systems for mass surveillance of Americans would violate fundamental rights.
The designation carries immediate business consequences. It restricts use of technology that a Pentagon source said had been applied in military operations in Iran, and it places a cloud over Anthropic's ability to do future work with federal agencies. The company warned that the decision, if allowed to stand, could set a troubling precedent for other firms that negotiate operational limits with the government.
Anthropic's chief executive, Dario Amodei, has sought to clarify the practical scope of the designation, saying it is "a narrow scope" and that businesses could continue to employ Anthropic's tools for projects unrelated to the Defense Department. Nonetheless, the designation prompted a strong reaction from the White House and the Pentagon.
President Donald Trump directed federal agencies to end contracts with Anthropic, and both he and Secretary Hegseth said the government would phase out its work with the company over a six-month period. Investors in Anthropic moved quickly to assess and contain the commercial fallout from the split with the Pentagon.
The chain of events traces back to negotiations that stretched over several months and culminated in actions taken by the White House and Defense Department on February 27. Those actions came shortly after a meeting between Amodei and Hegseth in which the company reportedly sought a resolution.
Anthropic also addressed an internal memo that became public earlier in the week. The memo, written last Friday and later published by The Information, included a remark from Amodei suggesting that some Pentagon officials' displeasure with the company stemmed in part from political differences: "we haven’t given dictator-style praise to Trump." Amodei apologized for the memo as he reiterated the company's intention to pursue legal remedies.
The Defense Department has signed agreements in the past year with several major AI labs, with contract values reaching up to $200 million each. Those accords include deals with Anthropic as well as other large AI companies. Following the Pentagon's move to blacklist Anthropic, Microsoft-backed OpenAI announced a separate agreement to make its technology available on the Defense Department network. OpenAI's CEO, Sam Altman, said the Defense Department shared the company's principles on maintaining human oversight of weapon systems and opposing mass U.S. surveillance.
With litigation now underway, the dispute between Anthropic and the Defense Department will play out in federal court. The outcome could determine not only Anthropic's immediate government business but also the broader contours of how AI firms and the U.S. military negotiate limits on technology use.
Summary
Anthropic filed a lawsuit in federal court to overturn a Pentagon national security supply-chain risk designation after refusing to remove guardrails preventing its AI from use in autonomous weapons and domestic surveillance. The move led the White House to direct a six-month phase-out of government work with Anthropic and prompted investor concern. The case raises legal and policy questions about the balance between national defense prerogatives and corporate limits on technology deployment.