US Appeals Court Upholds Pentagon Blacklisting of Anthropic as Supply Chain Risk

A federal appeals court in Washington, D.C., has upheld the Pentagon’s designation of Anthropic as a supply chain risk, rejecting the AI company’s legal challenge against the Trump administration. The 2-1 decision by the U.S. Court of Appeals for the District of Columbia affirmed that the Department of Defense (DOD) had sufficient grounds to maintain the ban on Anthropic’s Claude models from military systems and defense contractor workflows.

Legal Ruling and Dissent

Judge Gregory Katsas, writing for the majority, stated that the DOD had "ample support" for concluding that integrating Claude into its information systems presented a statutorily covered national-security risk. The majority opinion argued that the Department reasonably feared Anthropic might manipulate Claude’s design to prevent it from performing contractually authorized national-security functions. Judge Karen LeCraft Henderson dissented from the ruling.

Background of the Dispute

The conflict originated in March, when the DOD labeled Anthropic a supply chain risk after negotiations over deploying Claude on the GenAI.mil platform collapsed. Anthropic had previously held a $200 million contract with the Pentagon signed in July 2025. The breakdown occurred because the DOD demanded unfettered access to the models for all lawful purposes, while Anthropic sought assurances that its technology would not be used for fully autonomous weapons or domestic mass surveillance.

Next Steps and Context

Anthropic has stated it remains confident in its position and is considering further review, including petitions for rehearing or Supreme Court intervention. The appellate panel delayed the decision’s immediate effect to allow for these legal options. This ruling follows a separate San Francisco federal judge’s decision last month that found one of the DOD’s parallel designations illegal, highlighting a split in judicial outcomes regarding the government's actions against the AI provider.