The U.S. Court of Appeals for the District of Columbia Circuit, in a 2-1 decision issued Friday, found that the Pentagon had "ample support" to conclude that integrating Anthropic's Claude models into its information systems presented a national security risk. This ruling allows the Defense Department to exclude Anthropic from its supply chain under federal procurement law. The court's majority opinion, written by Judge Gregory Katsas, emphasized the deference courts give to national-security judgments.
The dispute originated in February when Anthropic declined the Pentagon's demand to remove restrictions on Claude's use for fully autonomous lethal weapons and mass domestic surveillance of Americans. The Pentagon, through its Secretary Pete Hegseth, subsequently designated Anthropic as a "supply-chain risk" in March, leading to the cancellation of existing military contracts and barring other defense contractors from using Anthropic's technology for military projects.
According to court documents, an Anthropic executive questioned the use of Claude by Palantir, a Pentagon contractor, during the capture of Venezuelan President Nicolás Maduro in January. This objection reportedly "led to alarm" within the Defense Department, raising doubts about whether Anthropic's software might cease functioning or cause other operational issues that could endanger military personnel. Under Secretary of Defense Emil Michael was quoted in the opinion as expressing these concerns.
Anthropic argued that it could not modify its models once delivered to the military. However, the appeals court determined that Anthropic controls the behavior of each new version, and the military must stay updated on the latest offerings. The court's decision focused on restrictions already embedded during training and those Anthropic could implement in future releases.
This ruling from the D.C. Circuit Court of Appeals conflicts with a separate decision from a federal court in San Francisco last month. The San Francisco court had ruled that a broader ban imposed by the Trump administration against Anthropic was unlawful, finding it constituted unlawful retaliation in violation of the First Amendment. Because the Pentagon based its blacklisting on two distinct legal designations, separate litigation was required in different courts.
Anthropic stated that it respectfully disagrees with the D.C. Circuit's decision and is considering all options, including further review. The company maintains that another federal court has already deemed a parallel government designation unlawful. Following the Pentagon's action, OpenAI, a competitor to Anthropic, reached its own agreement with the Defense Department.