Anthropic asks D.C. Circuit to rehear supply-chain risk case

On Oct. 13, Anthropic asked the U.S. Court of Appeals for the D.C. Circuit to rehear its decision upholding the Pentagon’s designation of the AI company as a national-security supply-chain risk.

The filing seeks either panel rehearing, which would return the case to the three judges who decided it, or rehearing en banc by the broader appeals court. It is a request for reconsideration, not a new ruling, and does not itself reverse the court’s Sept. 25 judgment. That 2–1 decision left in place a Pentagon action requiring Claude’s removal from department systems and restricting contractors from using Anthropic products for Pentagon work.

Anthropic argues in the petition that the panel majority stretched a law aimed at hostile foreign actors into one that presumptively covers every government contractor. The company also argues that the ruling allowed the Pentagon to act without a pre-designation process despite what Anthropic calls evidence that the designation retaliated against its speech about AI safety. Those are Anthropic’s legal claims in seeking rehearing; the court has not adopted them.

In the Sept. 25 opinion, Judges Gregory Katsas and Neomi Rao denied Anthropic’s consolidated petitions for review. The majority said the Pentagon had ample support for finding that continued integration of Claude into department or contractor systems created a risk covered by the Federal Acquisition Supply Chain Security Act. It cited Anthropic’s ability to encode restrictions into Claude and instances when those restrictions prevented the model from performing tasks requested by government users.

The majority also rejected Anthropic’s procedural and First Amendment claims. It said the department promptly notified Anthropic and gave the company a fair opportunity to challenge the exclusion. On the speech claim, the court concluded that the Pentagon acted because Anthropic would not accept an “all lawful uses” contract term, rather than because the company advocated for AI regulation.

Judge Karen LeCraft Henderson dissented. She argued that the statute’s phrase “otherwise manipulate” should be read narrowly to require intentionally subversive or deceptive conduct and that Anthropic therefore fell outside the statutory definition. The panel ruling applies to the Pentagon’s supply chain; it does not impose a government-wide ban on Anthropic products.