WASHINGTON (Realist English). On September 25, the US Court of Appeals for the District of Columbia issued a ruling refusing to block the Department of Defense’s order to blacklist the company Anthropic.
This decision left in force the Pentagon’s classification of Anthropic as a “national security supply chain risk,” meaning that US military contractors cannot use Anthropic’s Claude model when working for the Department of Defense.
Basis of the Ruling: Model Restrictions Became the Key Factor
In its ruling, the appellate court established that Anthropic’s very ability to restrict the behavior of its model makes the Pentagon’s ban lawful. Judge Gregory Katsas, in the majority opinion, wrote: “Anthropic acknowledges that the company encoded restrictions in Claude that prevent the model from performing tasks that Anthropic does not wish it to perform. These restrictions have repeatedly prevented the performance of tasks requested by government users.”
Katsas further noted that a dispute had previously arisen over whether the contract’s prohibitive provisions blocked the use of Claude during ongoing overseas military operations, “which deprived the Department of Defense of the ability to determine whether Claude would perform tasks as needed and as expected.”
“In our republic, the president and the secretary of defense must decide how best to balance these competing risks,” Katsas wrote. Judge Karen Henderson issued a dissenting opinion, stating that this decision means that any supplier “either accepts the secretary’s demands or risks being deemed a threat to national security.”
Background of the Conflict: The Dispute Over Guardrails
The conflict began in February 2026. Anthropic refused to comply with Secretary of Defense Pete Hegseth’s demand to remove restrictions on the use of its Claude model in fully autonomous lethal weapons and mass domestic surveillance of Americans.
Anthropic CEO Dario Amodei insisted on maintaining guardrails prohibiting the use of the technology for fully autonomous weapons and mass domestic surveillance. In March, the Pentagon officially designated Anthropic as a “supply chain risk,” canceled its military contracts, and banned other Department of Defense contractors from using its technology.
This designation is highly unusual — previously such authority was applied only to foreign companies associated with hostile forces.
Legal Conflict: Disagreement Between California and Washington
The Washington appellate court’s ruling enters into a potential legal conflict with a previous ruling by a federal court in California. In late August, federal judge Rita Lin in San Francisco ruled that the Trump administration’s broader sanctions against Anthropic (including an order for all federal agencies to stop using Claude) were unlawful.
In a 59-page ruling, Judge Lin found that the sanctions “lack a legal basis and violate constitutional rights,” and criticized the government for “empty invocation of national security to punish critics of the government.” She noted that the measure appeared to be aimed at punishing Anthropic for public criticism, which violates the First Amendment.
However, the Pentagon ban upheld by the Washington appellate court is based on federal procurement rules, not the law on which the California case was based, so both rulings can exist in parallel. The Pentagon ban currently remains in force.
Anthropic’s Response and Prospects
A representative of Anthropic told US media: “We respect but disagree with the court’s decision. Another federal court has already found the government’s parallel order unlawful. We are confident in our position and are considering all options, including further judicial review.”
This ruling has dealt a serious blow to Anthropic. The company is valued at nearly $1 trillion and is preparing for a long-awaited IPO in the fall. The Pentagon’s designation has already led to the loss of billions of dollars in business and undermined the company’s reputation ahead of the IPO.
According to reports, the US military had previously used Claude in several classified and sensitive systems, including the January operation to capture Venezuelan leader Maduro. Since relations with the White House deteriorated, nearly three-week export restrictions were imposed on Anthropic’s most advanced model in June, blocking use by all non-American clients.
This legal war is unfolding against the backdrop of Trump’s statement about wanting to maintain US leadership in AI over China. The White House is displeased with Amodei’s repeated warnings about the dangers of AI, which Trump has called a “hoax.”
