A federal appeals court in Washington, D.C., on Wednesday rejected Anthropic’s bid to pause Pentagon actions that the company says amount to a blacklist and could affect its ability to work with the U.S. government, even as Anthropic continues to press its legal challenge. The decision came in the latest round of a dispute that Anthropic has framed as retaliation tied to how its Claude chatbot might be deployed in defense systems.
In its filing, Anthropic asked the U.S. Court of Appeals for an order that would shield the San Francisco company from what it described as the fallout from a Pentagon conflict over the use of its technology in fully autonomous weapons and in potential surveillance of Americans while the appeals panel collects evidence. The court denied that request in Washington, even though Anthropic had won in a separate case in San Francisco that addressed the same overall issues.
The appeals court’s setback followed a San Francisco ruling by U.S. District Judge Rita Lin that the Trump administration had overstepped by labeling Anthropic a supply chain or security risk that was not qualified for military contracting, along with issuing other directives that the company said could cripple it as it competes for AI primacy against rivals such as OpenAI and Google. After that San Francisco decision, the Trump administration removed the stigmatizing labels and took steps that, according to a court filing made in San Francisco earlier this week, cleared the way for government employees and contractors to continue using Claude and other chatbots.
Anthropic’s separate lawsuits were filed last month, one in San Francisco and one in Washington. In both, the company asserted that the Trump administration was engaging in what Anthropic described as an “unlawful campaign of retaliation” for its attempt to impose limits on how its AI technology could be deployed.
The Pentagon’s position, as described by the company’s filings and arguments, drew a sharp response from the administration, which blasted Anthropic as a liberal-leaning company seeking to dictate U.S. military policy. In Washington, the appeals court acknowledged that Anthropic would “likely suffer some degree of irreparable harm” if it were treated as a supply chain risk, but it concluded that the record did not yet justify the court issuing the kind of immediate order Anthropic sought. The appeals court said the “precise amount of Anthropic’s financial harm is not fully clear,” a consideration that helped drive the panel’s refusal to overturn the administration’s actions on an emergency basis.
The case in Washington is not finished. Further evidence is scheduled to be presented at a hearing set for May 19, with the appeals panel continuing to weigh the parties’ competing arguments about the legality and scope of the Pentagon’s designations.
Anthropic said it was encouraged by what it characterized as the court’s recognition that the issues needed to be resolved quickly, and it said it remained confident the courts would ultimately agree that the supply chain designations were unlawful. Matt Schruers, the CEO of the technology trade group Computer & Communications Industry Association, said the conflicting rulings so far could create significant business uncertainty for AI companies at a pivotal time.