A federal appeals panel in Washington, D.C. on Wednesday rejected Anthropic’s request to temporarily block the Defense Department’s national security designation that prevents the company from obtaining Pentagon contracts. The court’s decision keeps in place a label that could broaden into government-wide restrictions pending further legal proceedings.
The action represents a partial legal victory for the Trump administration and contrasts with a separate outcome in California, where a federal judge has already enjoined one of the Pentagon’s orders. The D.C. panel’s refusal to pause the designation is not a final adjudication of the underlying dispute.
Anthropic, the developer of the Claude AI assistant, has challenged Defense Secretary Pete Hegseth’s decision to designate the company as a supply-chain risk. That designation bars Anthropic from participating in Pentagon contracts and may trigger additional blacklisting across civilian agencies after an interagency review described in the D.C. litigation.
Company executives contend the designation could inflict substantial financial damage and reputational harm, potentially costing Anthropic billions in lost business. In its legal filings, Anthropic says the Pentagon’s action amounted to retaliation for the company’s refusal to permit the military to use Claude for U.S. surveillance or autonomous weapons programs, citing safety and ethical concerns.
The current D.C. appeal concerns a statute that could widen the scope of the blacklist to the broader civilian government through an interagency process. Anthropic has pursued two separate lawsuits challenging two distinct orders issued by Secretary Hegseth under different procurement statutes. The California lawsuit dealt with a narrower statute that excludes the company from Pentagon contracts tied to military information systems.
On March 26, a California federal judge temporarily blocked one of Hegseth’s orders, finding that the Pentagon appeared to have unlawfully retaliated against Anthropic for its views on AI safety. That ruling stands apart from the D.C. panel’s denial of a stay in the second, broader challenge.
In its complaints, Anthropic argues the government violated constitutional protections. The company says the designation penalized its protected speech under the First Amendment and deprived it of a chance to contest the label in violation of its Fifth Amendment due process rights. Anthropic also contends the designations lack factual support and are inconsistent with previous praise the military has given to Claude.
The Justice Department has countered that Anthropic’s refusal to accept contractual terms created operational uncertainty for the Pentagon and could jeopardize military systems during operations. In court filings, the government framed its action as a response to the company’s contractual stance rather than to its public statements on AI safety.
The D.C. litigation centers on the potential for the Pentagon’s decision to lead to a wider blacklist affecting civilian agencies after the prescribed interagency review. The California case remains focused on a narrower exclusion tied specifically to military information systems contracts.
Context and next steps
Both cases will continue to move through the federal courts. The D.C. panel’s denial of a pause preserves the Pentagon’s designation in that jurisdiction while the legal process unfolds. The differing outcomes in D.C. and California mean that the ultimate scope and duration of the restrictions will depend on further rulings and the resolution of constitutional challenges.