Appeals court backs Pentagon’s Anthropic risk label, while California ruling still stands

A federal appeals court has upheld the Pentagon’s designation of Anthropic as a supply-chain risk, but a separate California order limiting the government’s ban remains in force for now. The split leaves a major AI-company dispute unresolved on two legal fronts.

Anthropic Watch analysis

What happened

Bloomberg reports that the US Court of Appeals for the DC Circuit ruled 2–1 against Anthropic’s challenge to the designation. The majority said the Pentagon had sufficient support under the relevant law to conclude that continued integration of Claude into government systems presented a national-security risk.

Judge Gregory Katsas wrote that Anthropic had refused to relax its restrictions on using Claude for lethal autonomous warfare or domestic surveillance. The court said that a separate California judge’s ruling remained in effect: that judge had found the government’s ban was not adequately justified under a different, narrower legal definition of supply-chain risk. Anthropic said it respectfully disagreed and was considering further review.

Why it matters

The ruling does not settle whether the government can bar Anthropic’s technology from federal contracts. It does, however, uphold the specific supply-chain-risk designation in the Washington case, while the California order continues to constrain a broader ban. The distinction matters to agencies, contractors and other AI firms weighing how much control a company can retain over military use of its models.

The dispute also puts two questions in sharper focus: how far national-security powers reach, and whether AI suppliers can set limits on uses they consider unacceptable. Those are legal and policy questions, not merely a contract negotiation with unusually large consequences.

Our read

This is a consequential setback for Anthropic, but not a clean win for the administration: the parallel California ruling remains in place. Keep the two cases separate, and the status of each clear. Court decisions are not interchangeable just because they involve the same company and a very busy set of lawyers.

What to watch

• Whether Anthropic seeks further review of the DC Circuit decision.
• Whether the government appeals or otherwise challenges the California order.
• How the rulings affect federal agencies’ and contractors’ use of Claude.
• Whether the dispute changes how AI companies write limits on military or surveillance uses.

Discussion spark: Should AI companies be able to set binding limits on military uses of their models, or should those decisions rest with elected governments?

Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.