A federal appeals court handed the Pentagon a major win Friday, ruling 2-1 that the Defense Department can continue treating AI company Anthropic as a supply chain risk and purging its Claude models from military systems. The decision from the U.S. Court of Appeals for the D.C. Circuit lands squarely against Anthropic in one of two parallel legal battles over the company's national security designation, even as a separate federal judge in California has already declared the same blacklisting unlawful.
The split rulings leave Anthropic in legal limbo at a moment the company says it can least afford it: billions of dollars in lost business are piling up, and a highly anticipated initial public offering hangs in the balance.
Judge Gregory G. Katsas, writing for the majority, framed the dispute as a straightforward national security judgment call. The court found the Defense Department had legitimate grounds to worry that Anthropic could alter Claude's design in ways that would block the military from using the AI system for functions the Pentagon considers authorized and necessary.
Katsas wrote in the majority opinion:
"The Department reasonably feared that Anthropic might manipulate Claude's design to prevent it from performing national-security functions that the Department deems contractually authorized and necessary."
The ruling permits the Pentagon to keep removing Claude from its systems and to bar Anthropic's products from all Defense Department work. One judge dissented, though the New York Post reported no details about the dissenting opinion's reasoning.
Anthropic said it "respectfully disagrees" with the decision. A company spokesperson stated:
"Another federal court has already held the government's parallel designation unlawful. We remain confident in our position and are considering all options, including further review."
The White House did not immediately respond to a request for comment.
The conflict traces back to February, when President Donald Trump and Defense Secretary Pete Hegseth accused Anthropic of endangering national security and slapped the company with a supply chain risk designation. That label made Anthropic the first American company ever publicly branded under the obscure government procurement statute, as Reuters reported. The designation barred defense contractors from using Claude in any Pentagon-related work.
The rupture grew out of failed negotiations over how the military could deploy Claude. Anthropic CEO Dario Amodei refused to remove two internal safety guardrails: one prohibiting fully autonomous AI weapons control (requiring human operators in the loop) and another blocking mass domestic surveillance. Amodei's position was that the company's products were not reliable enough for those applications, and he declined to back down despite mounting pressure from the administration.
The Pentagon had awarded Anthropic a $200 million contract in July 2025, Fox News reported, but threatened to terminate the deal if Claude was not approved for all lawful uses. Claude was, at the time, the only commercial AI system approved for classified military use and was deeply embedded in military workflows. When Anthropic held firm, the administration struck a deal with OpenAI to replace Claude with ChatGPT, National Review reported.
Hegseth framed the dispute as a matter of military readiness. He stated that the government could not allow the U.S. military to be constrained by a private company's internal policies on how its technology could be used, Just The News reported.
The D.C. Circuit's ruling stands in direct tension with a decision issued this summer by U.S. District Judge Rita Lin of the Northern District of California. Judge Lin found the Pentagon acted illegally by punishing Anthropic for its public criticism of the administration's views on AI use. Her 59-page order called the supply chain risk designation "illegal and baseless."
Lin's opinion went further than simply finding a procedural flaw. She concluded that the supply chain risk statute was never intended to apply to domestic companies, that the administration bypassed required procedural safeguards, and that targeting Amodei for his public statements constituted illegal First Amendment retaliation. Her language was pointed:
"Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government."
She also wrote that the "empty invocation of national security is not a blank check to punish and retaliate against government critics," as Breitbart reported.
The California ruling prompted sharp pushback from the Pentagon. CTO Emil Michael called it "a disgrace" filled with "dozens of factual errors" that seeks, "DURING A TIME OF CONFLICT," to "upend the [president's] role as Commander in Chief."
The two cases targeted different legal theories. The California case challenged the constitutionality and statutory basis of the supply chain risk designation itself, with Anthropic alleging violations of its First Amendment free speech rights and Fifth Amendment due process rights. The D.C. Circuit case focused on a different rule the Pentagon invoked to declare Anthropic a supply chain risk. Authorities have not publicly clarified whether the D.C. Circuit ruling affects the California injunction, or whether the two operate on independent legal tracks.
The practical result, for now, is a split. One federal court has ordered the Pentagon to stop enforcing the blacklisting. Another has said the Pentagon can keep going. Anthropic has signaled it may seek further review of the D.C. ruling, though the company has not specified whether that means en banc rehearing before the full D.C. Circuit or a petition to the Supreme Court.
Anthropic has stated that the blacklisting has cost it billions of dollars in lost business and damaged its reputation ahead of a planned initial public offering. The company's characterization of the financial damage has not been independently verified in public filings or court records described in available reporting. But the scale of the claim reflects how deeply Claude was woven into defense work before the designation landed.
The loss is not just financial. Being labeled a supply chain risk carries a reputational stigma that extends well beyond Pentagon contracts. Defense contractors who might otherwise partner with Anthropic face their own compliance risks if they use products from a company the government has formally flagged. That chilling effect ripples outward through the defense industrial base, regardless of what any single court has ruled.
Investigators and courts will need to resolve several open questions before this fight reaches any final resolution. The identity and reasoning of the D.C. Circuit's dissenting judge remain undisclosed in public reporting. The specific procurement rule at issue in the D.C. case has not been publicly detailed. And the broader constitutional question raised by Judge Lin in California, whether the government can use national security procurement statutes to punish a domestic company for speech, has not been tested at the appellate level on the West Coast.
The administration has maintained its position that the supply chain risk designation is a legitimate exercise of executive authority over military procurement. Anthropic has maintained that the designation is retaliation dressed up as national security. Both sides now have a federal court ruling in their pocket.
When two federal courts look at the same government action and reach opposite conclusions, the question stops being whether someone is right. It becomes a question of which branch of government gets the final word on where national security ends and constitutional rights begin.
