Full Breakdown
Pentagon Ends Use of Anthropic’s Claude AI After Supply-Chain Risk Designation
By Drooid · · How we work
Pentagon Halts Use of Anthropic’s Claude AI
A senior Department of Defence official announced that the Pentagon has stopped employing any Anthropic products, including the Claude model, for research, analysis, intelligence gathering and military operations. The cessation follows Secretary of Defence Pete Hegseth’s February label of Anthropic as a “national-security supply-chain risk,” a designation that required the department to end use by late August. The statement did not explain the delay in fully implementing the ban.
Background and Supply-Chain Risk Designation
Anthropic’s Claude had been used by U.S. government agencies since 2024 and was the first advanced AI system cleared for classified work. In February, Hegseth gave Anthropic three days to remove safety guardrails that prevented fully autonomous lethal-weapon functions and mass-surveillance capabilities. Anthropic refused, citing ethical concerns, prompting the February designation that barred the company from military contracts and prohibited defense contractors from integrating its models. Anthropic later described the move as “unprecedented and unlawful” and filed a lawsuit seeking to overturn the label.
Court Rulings and Ongoing Legal Fight
A federal appeals court upheld the Pentagon’s designation in a September 25 decision, with Judges Gregory Katsas and Neomi Rao forming the majority and Judge Karen LeCraft Henderson dissenting. Earlier, in August, U.S. District Judge Rita Lin ruled the Pentagon had illegally retaliated against Anthropic and violated due-process rights. The appeals panel’s reversal left the supply-chain risk label in place, and Anthropic is considering further appeal to the full circuit or the Supreme Court.
Federal Adoption Outside the Defense Department
Despite the defense ban, Anthropic released “Claude for Government” on September 30, offering the model in a FedRAMP High environment—the highest U.S. government cloud-security tier. Federal and state agencies can purchase the service directly, through Carahsoft, or via the GSA’s OneGov channel, with a promotional $1-per-user rate available through October 31. The restriction applies only to military systems and defense-contractor work, allowing broader government use while the Pentagon remains shut out.
Financial Stakes and Outlook
Anthropic signed a $200 million contract with the Pentagon in July 2025, and the loss of that business coincides with the company’s preparation for a pre-Thanksgiving IPO that could value it near $2 trillion. The firm reported a $42 billion net loss for 2025, alongside revenue that grew twelvefold to $4.6 billion. Claude’s popularity surged after the blacklist, reaching number one in U.S. app-store downloads in March. The outcome of the legal dispute will likely influence the IPO roadshow and Anthropic’s broader government strategy.