Anthropic wins court fight over Pentagon blacklist
US District Judge Rita Lin, sitting in California, ruled on Thursday that the Trump administration's decision to designate Anthropic a national security "supply-chain risk" amounted to unconstitutional retaliation. Her 59-page ruling vacated the February 27 decision by Defense Secretary Pete Hegseth that had made Anthropic ineligible for federal contracts. It also lifted a separate, additional measure Hegseth had imposed that barred contractors and suppliers to the US military from doing business with Anthropic; Lin called that measure "arbitrary, capricious, an abuse of discretion, and otherwise not in accordance with law." On the underlying designation itself, she wrote that while the Department of War "is undisputedly free to select the AI vendor of its choice, the evidence demonstrates that the broad measures imposed on Anthropic were illegal and baseless."
Lin also found that nine federal agencies, including the Pentagon, the Treasury Department, the State Department and the Department of Homeland Security, had improperly imposed sanctions on Anthropic tied to the same designation, and her ruling removes those penalties too. It stops short of forcing the Pentagon's hand going forward: Lin affirmed that the Pentagon is not required to use Anthropic's models and remains free to choose other vendors instead. A Pentagon spokesperson could not immediately be reached for comment, and the department is expected to appeal. Anthropic spokesperson Danielle Cohen said in a statement: "We welcome the court's ruling that this supply-chain risk designation was unlawful. We remain focused on working productively with the government to harness AI for our national security."
The case traces back to a dispute earlier this year over a $200 million Pentagon deal to use Anthropic's Claude models for military applications. The rift opened after reports that the US had used Claude in the operation to capture Venezuelan president Nicolás Maduro; afterward, a Palantir employee relayed concerns from an Anthropic staffer to US officials about how the models had been used. During the negotiations, Anthropic had insisted on certain limits on how its models could be deployed, including support for lethal autonomous weapons and mass surveillance systems. Hegseth rejected any curbs, arguing that a contractor could not dictate how the technology was used once deployed and insisting the contract allowed "all lawful use." After talks broke down in February, Hegseth was widely seen as punishing Anthropic by designating it a supply-chain risk, a national security label that blacklisted the company from doing business with the federal government. At the time, the Pentagon defended the label, saying that giving Anthropic access to any classified systems would "introduce unacceptable risk" if the company could disable or alter its technology, for instance during wartime.
Anthropic responded by filing two lawsuits accusing the Pentagon of violating its First and Fifth Amendment rights on ideological grounds: one in federal district court in California, the case decided by this ruling, and one at the US Court of Appeals for the District of Columbia, which is still ongoing. In her opinion, Lin pointed to the administration's own continued engagement with Anthropic as evidence against the government's stated rationale, noting that "even now, the government is discussing collaboration with Anthropic on its new model, Mythos, in an array of sensitive contexts," and that "none of that is consistent with a genuine fear that Anthropic is a saboteur who would poison its software to harm national security." Separately from the supply-chain risk fight, Anthropic's models have also recently become subject to the Trump administration's AI oversight framework because of their capabilities.
Key facts
- US District Judge Rita Lin ruled on Thursday that the Pentagon's February 27 designation of Anthropic as a national security "supply-chain risk" was unconstitutional retaliation, and vacated it in a 59-page ruling.
- The ruling also lifted a measure barring US military contractors and suppliers from doing business with Anthropic, and found that nine agencies, including the Pentagon, Treasury Department, State Department and Department of Homeland Security, had improperly sanctioned the company.
- The dispute stems from a $200 million Pentagon deal for Claude models that broke down after Anthropic sought limits on military deployment and Defense Secretary Pete Hegseth insisted the contract allowed "all lawful use."
- The ruling does not require the Pentagon to use Anthropic's models, and the department is expected to appeal; Anthropic still has a second, separate suit pending at the US Court of Appeals for the District of Columbia.
- Judge Lin cited the government's ongoing discussions with Anthropic about its new model, Mythos, as evidence against the Pentagon's stated rationale for the blacklist.
Why it matters
The ruling is a direct rebuke of how far a federal agency can go in punishing an AI vendor for pushing back on contract terms: Anthropic had tried to negotiate limits on how its models could be deployed, including around lethal autonomous weapons and mass surveillance, and the Pentagon's response was to declare the company a security risk and cut it off from federal business. A court calling that response "illegal and baseless" bears on how AI companies can negotiate deployment terms with the government going forward, at the same time as the Trump administration is separately expanding its oversight of powerful AI models, an effort that now also covers Anthropic's systems.
Who it affects
Anthropic regains standing to bid for federal contracts and to work with companies that supply or contract with the US military, and the nine agencies named in the ruling, among them the Pentagon, Treasury Department, State Department and Department of Homeland Security, have had their sanctions against the company voided. The ruling also matters to other AI vendors weighing whether they can set conditions on military use of their models without risking blacklisting, and to the Pentagon itself, which must now decide whether to appeal and how to handle its unresolved $200 million relationship with Anthropic.
How to use it
Practically, the ruling restores Anthropic's eligibility rather than settling the underlying contract dispute: the Pentagon is not obligated to buy Anthropic's models and can still choose other vendors, so nothing here guarantees Claude returns to military use. An appeal is expected, and a second Anthropic suit over the same conduct is still active at the US Court of Appeals for the District of Columbia, so whether Anthropic actually ends up doing business with the Pentagon again on terms it finds acceptable remains an open question.
How solid is it
The account rests directly on Judge Lin's 59-page written ruling and includes what read as verbatim excerpts from it, plus an on-the-record statement from Anthropic spokesperson Danielle Cohen, which is a solid basis for the core facts. The weaker points sit on the Pentagon's side: its spokesperson could not be reached, so the claim that the department will appeal is the reporting's own expectation rather than a confirmed Pentagon statement, and neither the exact grounds for any appeal nor the procedural status of the still-open DC Court of Appeals case beyond "ongoing" is specified. The description of what limits Anthropic sought in negotiations, including around lethal autonomous weapons and mass surveillance systems, is also loosely worded in the reporting and is best read as the general shape of the dispute rather than a precise contract term.
Risks and caveats
Winning this ruling does not resolve the fight: the Pentagon is expected to appeal, the ruling explicitly does not force the department to use Anthropic's models, and the DC Court of Appeals case over the same underlying conduct remains open. The core disagreement, whether Anthropic can attach conditions to military use of its models while the Pentagon insists on "all lawful use," is unresolved and could resurface in any renewed negotiation. Anthropic also remains under a separate track of government scrutiny: its models were recently brought under the Trump administration's AI oversight framework because of their capabilities, a development unrelated to this case but a reminder that the wider relationship between Anthropic and the US government stays contentious even after this win.
“Though the Department of War is undisputedly free to select the AI vendor of its choice, the evidence demonstrates that the broad measures imposed on Anthropic were illegal and baseless.”
— Judge Rita Lin, in her ruling