Federal appeals court lets Pentagon keep Anthropic on supply-chain risk list

The US Court of Appeals for the DC Circuit on Friday denied Anthropic's petition and left intact the Pentagon's designation of the company as a "supply-chain risk," a decision that allows the Defense Department to continue blocking Claude models from military and contractor systems. The panel ruled two-to-one that the department had an adequate basis to conclude that integrating Claude poses a security risk under the law, noting explicitly that the company itself acknowledges it codes restrictions preventing the model from performing certain tasks.

The dispute began earlier this year when Defense Secretary Pete Hegseth issued two separate classification orders against Anthropic under different supply-chain statutes, aiming to keep the models out of the federal government through this month. The stated reason: Anthropic's refusal to allow deployment of its current models for autonomous weapons or internal surveillance, a stance Hegseth called a significant security risk. The two lawsuits split across courts: a federal judge in San Francisco vacated one order in March and affirmed that ruling last month, but Friday's decision leaves the second order in effect indefinitely, pending final appeals that could take years.

The majority wrote that the Pentagon acted within proper procedure and that the dispute is essentially a standard contract negotiation, not punishment for a political stance. "The department excluded Anthropic from the supply chain because it refused to agree to a contract term the department deemed essential, not because of its support for government regulation of artificial intelligence," the majority opinion stated. The company's due-process and free-speech claims were rejected. As early as April the same panel declined to stay the order temporarily, finding Anthropic had not met the stringent requirements for preliminary relief; oral argument revealed a divided panel, and the final decision came down to a single vote.

Anthropic reported lost revenue immediately after the classification as customers shied away from working with a "government pariah," though the company has provided no quantitative update since and continues to cite growing sales ahead of an expected public offering later this year. The Pentagon has not detailed what stage it has reached in replacing Claude with alternatives such as SpaceX's Grok, Google's Gemini, or OpenAI's GPT models. Employees at Google and OpenAI protested the military cooperation that Anthropic declined, but leadership at both companies dismissed the protests and called support for the US government essential.

Anthropic spokesperson Danielle Cohen said the company remains confident in its position and is evaluating all options, including an appeal to the full appeals court or to the Supreme Court. Both orders are expected to shadow the company through protracted litigation, but for now the Pentagon and much of the Trump administration may continue to steer clear of Claude, just as Anthropic prepares for its initial public offering.