A federal appeals court has sided with Defense Secretary Pete Hegseth in his fight with AI company Anthropic, ruling 2-1 that his decision to cut the firm’s Claude model out of the Pentagon’s supply chain fell within his legal authority. The U.S. Court of Appeals for the District of Columbia Circuit handed down the decision Friday, with the majority calling Hegseth’s worries about the technology “deeply sobering.”
The split had been telegraphed back in May, when the court heard oral arguments. U.S. Circuit Judges Gregory Katsas and Neomi Rao, both Trump appointees, appeared inclined to back the administration, while Judge Karen LeCraft Henderson, appointed by George H.W. Bush, described Hegseth’s move at the time as “just a spectacular overreach by the department.” That lineup held: Katsas wrote the majority opinion, Rao joined it, and Henderson dissented alone.
Katsas didn’t shy away from the stakes involved. “This case raises profoundly difficult questions about the appropriate military uses of an almost unimaginably powerful new technology,” he wrote, noting Hegseth’s fear that AI models with too many restrictions could shut down mid-operation and cause missions to fail. At the same time, he acknowledged Anthropic’s own warning — that AI systems without enough guardrails risk “hallucinating inappropriate targets for lethal military force.”
“Both possibilities present obvious national-security concerns,” Katsas and Rao wrote. “But in our Republic, it is the President and the Secretary of War who must determine how best to balance the competing risks.” The judges concluded that in making that call, Hegseth “did not transgress any limits on his authority under the Supply Chain Security Act or the Constitution.”
Government lawyers had argued that whatever procedural missteps occurred, they were harmless, and that Hegseth’s decision reflected legitimate “considered national-security judgment” deserving of judicial deference. Anthropic countered that the whole episode amounted to “textbook retaliation,” dressed up with false justifications — in the company’s words, “a supply-chain risk designation in search of a justification.”
The dispute traces back to late February, when Trump and Hegseth publicly moved to cut ties with the AI firm, branding Anthropic “WOKE” and “sanctimonious” for refusing to let Claude be used for autonomous lethal warfare or mass surveillance of Americans.
The clash escalated after Anthropic CEO Dario Amodei defended the company’s stance, arguing that those two limits on Claude’s use were matters of “existential” importance. Trump responded by ordering every federal agency to immediately stop using Anthropic’s technology. Within days, the Pentagon formally labeled Anthropic a “supply-chain risk and threat to national security.”
The majority found nothing wrong with that reasoning, writing that the record “amply supports” Hegseth’s conclusion that he “reasonably feared” Anthropic could alter Claude’s design in ways that would block it from performing national-security functions the Department considers authorized and necessary under its contracts.
“We reject these challenges,” the majority wrote. “The Department had ample support for its conclusion that the continued integration of Claude into the Department’s information systems, by the Department or its contractors, presented a statutorily covered national-security risk.” The judges pointed out that Anthropic itself acknowledges building restrictions into Claude to block certain tasks the company objects to.
Henderson wasn’t persuaded. In her dissent, she argued the Federal Acquisition Supply Chain Security Act simply doesn’t give Hegseth the power he claimed. “The Department here made good on its promise to designate Anthropic a supply chain risk after Anthropic declined the Secretary’s ultimatum to replace its use restrictions on Claude with a general provision permitting ‘all lawful uses,'” she wrote.
She warned the ruling sets a troubling precedent for whichever company might eventually replace Anthropic as a Pentagon contractor. “Suppose the Secretary tells Anthropic’s presumed replacement to change its AI-use policies to permit any ‘functions that the Department deems necessary’ or it will share the same fate as Anthropic,” she wrote. “According to today’s decision, that contractor will have a choice: Agree to the Secretary’s demands or risk being designated a national security threat under FASCSA. I cannot agree that this is the scenario the Congress had in mind when it enacted FASCSA.”
