What happened
On 25 September 2026, a divided panel of the US Court of Appeals for the DC Circuit (2-1, Judges Katsas and Rao majority; Judge Henderson dissenting) upheld DoD's March 2026 designation of Anthropic as a supply chain risk under the Supply Chain Security Act, rejecting claims the blacklisting was arbitrary, retaliatory, or unconstitutional. The ruling blocks Anthropic's Claude from US military systems and from defense-contractor work for DoD. The court stayed immediate effect to allow rehearing/en banc or Supreme Court review, and a separate parallel designation was previously held unlawful in the Northern District of California, which remains in effect.
Why it matters
It is binding national-security action controlling how a leading US frontier AI model may be accessed and distributed to government customers. It also creates precedent for classification of AI labs as supply-chain risks and shows the legal system deferring to executive national-security determinations over AI firms' commercial/contract interests.
Action needed
Defense-contractor AI deployers must continue excluding Anthropic models from DoD work; other frontier labs should review their government-deployment restrictions and export/national-security exposure. Anthropic is considering further review.