In a major legal setback for artificial intelligence developer Anthropic, a federal appeals court in Washington, DC, has upheld the Department of Defense's decision to classify the firm as a supply-chain risk. The 2-1 ruling allows the US military and federal agencies to continue removing Anthropic's Claude AI models from their information systems. The conflict originated from Anthropic's strict policy prohibiting its AI technology from being deployed in autonomous weapons development or domestic surveillance, a position that Secretary of Defense Pete Hegseth identified as a significant national security risk.
Court Upholds Pentagon Risk Label Over Safety Controls
The majority opinion from the DC Circuit Court of Appeals stated that the defense department provided ample justification for its safety assessment regarding Claude's integration into federal infrastructure. The judges highlighted that Anthropic explicitly encodes restrictions into its models to prevent tasks the company wishes to restrict. Dismissing Anthropic's claims that its due process and free speech rights were violated, the court characterized the action as a standard contractual disagreement. The majority noted that the Pentagon excluded Anthropic due to its refusal to assent to an essential contract term, rather than penalizing the company for advocating AI regulation.
Following the decision, Anthropic spokesperson Danielle Cohen stated that the company remains confident in its position and is evaluating all legal avenues. These options include seeking a rehearing before a broader panel of the DC Circuit Court of Appeals or escalating the dispute to the US Supreme Court. The latest ruling follows an April decision by the same panel, which declined to grant a temporary stay after finding that Anthropic failed to meet the stringent requirements necessary for immediate relief. While the three-judge panel appeared divided during oral arguments, the final judgment was delivered 2-1.
Divided Legal Battle Across Federal Courts
The sanctions against Anthropic were imposed under two separate supply-chain laws, requiring the company to challenge them in different jurisdictions. While Friday's appellate decision keeps one restriction active indefinitely, Anthropic previously secured a favorable outcome in San Francisco. A federal judge there struck down the other supply-chain risk label in March and reaffirmed that ruling last month. With both matters facing prolonged appeals, the legal battle between Anthropic and the government could take years to fully resolve.
Nevertheless, the decision in Washington allows the Pentagon and the Trump administration to proceed with phasing out Claude AI software by this month. Anthropic previously acknowledged that the initial government designation damaged its revenue as corporate customers grew hesitant to engage with a blacklisted entity. However, the company has reported growing sales in recent months and continues to move toward a potential initial public offering (IPO) later this year.
Rival AI Firms Step In Ahead of Planned IPO
With Claude being phased out, the Pentagon has turned to alternative AI solutions, including SpaceX's Grok, Google's Gemini, and OpenAI's GPT models. These government partnerships have sparked internal debate within competing tech companies. Certain employees at Google and OpenAI voiced objections to their employers securing military agreements that Anthropic rejected on ethical grounds. However, executive leadership at both Google and OpenAI brushed aside internal protests, emphasizing that supporting the US government remains a critical priority.
