A panel of the D.C. Circuit Court of Appeals ruled that the Pentagon may continue to treat Anthropic as a supply-chain risk, meaning that defense employees and their contractors are prohibited from employing the firm’s artificial-intelligence products in official work. The decision follows a February ultimatum from the Defense Department that the company declined to meet.
The designation was first announced in early March after the department sought to modify its contract with Anthropic, replacing earlier bans on domestic mass-surveillance and fully autonomous weapons with a clause allowing all lawful uses. The move marked a shift from a narrow restriction to a broader prohibition on any DOD-related activity involving the firm’s tools.
Anthropic had previously succeeded in overturning one of the department’s statutory justifications in an August decision by the Northern District of California. However, the company failed to overturn the second justification when the case reached the D.C. Circuit, where the appellate judges upheld the risk label on procedural and substantive grounds.
The majority opinion was authored by Judges Gregory Katsas and Neomi Rao, both appointed during the first Trump administration. A dissent was filed by Judge Karen LeCraft Henderson, who received her appointment from the first President Bush, highlighting a split view on the breadth of the department’s authority in this context.
Analysts expect the ruling to keep the Pentagon and its subcontractors from integrating Anthropic’s Claude models into their workflows, limiting the company’s exposure to defense contracts. Nonetheless, Anthropic’s overall revenue continues to expand rapidly, with growth rates reported as roughly tenfold each year, underscoring a divergent trajectory between government restrictions and commercial momentum.
Despite the setback, Anthropic is preparing for an initial public offering slated for November, with market observers projecting a valuation that could approach two trillion dollars. The contrast between the firm’s soaring market expectations and the ongoing security concerns illustrates the complex landscape facing AI developers that operate at the intersection of commercial ambition and national-security policy.