A federal appeals court has upheld the Pentagon’s designation of Anthropic, an artificial intelligence laboratory, as a supply chain risk, permitting the Department of Defense to continue removing the company’s AI models from its systems and restricting their use in defense-related work. The decision was issued on Friday by the U.S. Court of Appeals for the District of Columbia Circuit, which ruled 2-1 in favor of the Pentagon.
The dispute began earlier this year when former President Donald Trump and then-Defense Secretary Pete Hegseth publicly accused Anthropic of posing a national security threat, prompting the Pentagon to label the company a supply chain risk. Anthropic’s CEO, Dario Amodei, contested the government’s position, highlighting concerns that the Pentagon’s restrictions could impede the company’s efforts to prevent misuse of its AI products, including for mass surveillance or autonomous weaponry.
In the court’s majority opinion, Judge Gregory G. Katsas wrote that the Pentagon had substantial justification for its actions, citing Anthropic’s own admission that the company implements built-in limitations in its Claude AI models to restrict certain functionalities. These safeguards, the court noted, raised concerns within the department that Anthropic might deliberately configure its software to block national-security-related applications deemed essential by the Defense Department.
Judges Katsas and Neomi Rao, both appointees of former President Trump, formed the majority, while the dissenting opinion came from Judge Karen LeCraft Henderson, a nominee of former President George H. W. Bush. The judges emphasized that the designation hinges on the company’s actions rather than its intentions, acknowledging that Anthropic’s restrictions on Claude stem from concerns about privacy and AI safety. “We have no reason to doubt” the company’s “noble intentions,” the opinion stated.
Anthropic responded to the ruling by expressing disagreement and pointing to an earlier, separate federal court case in California that found the Pentagon’s parallel supply chain risk designation unlawful. The company said it remains confident in its position and is exploring further legal options.
The ruling does not extend to a broad government ban on Anthropic’s products but specifically allows the Department of Defense to exclude Claude from its operations and bar contractors from using Anthropic’s AI in Defense Department projects. The Pentagon began removing the models earlier this year following its initial designation.
The White House has yet to comment on the court’s decision. This case contrasts with a separate legal dispute pending in California regarding the Pentagon’s actions in response to Anthropic’s criticisms of military AI policies, which the appeals court noted without dispute.