Appeals Court Restores Pentagon Blacklist of Anthropic Over AI Military Limits

A U.S. federal appeals court has upheld the Defense Department's decision to blacklist artificial intelligence startup Anthropic from military contracts, reinstating the company's designation as a national security supply-chain risk, according to reporting from Tech Startups. The ruling reverses an August district court judgment that had struck down the designation as unlawful.

At the core of the dispute is whether commercial AI vendors can dictate operational limits when government agencies license frontier models. Anthropic restricted its Claude model from being deployed for mass domestic surveillance or lethal autonomous weapons systems without human oversight, citing inadequate testing for those high-stakes applications.

The conflict deepened following a February meeting between Defense Secretary Pete Hegseth and Anthropic chief executive Dario Amodei. Defense officials demanded that the startup relax its operational restrictions or face removal from military supply chains. President Donald Trump subsequently ordered federal agencies to cease using Anthropic tools, and the Defense Department formally classified the firm as a supply-chain risk—an authority typically aimed at foreign intelligence threats.

Following the order, agencies began severing operational ties. The State Department transitioned its internal StateChat platform from Anthropic to OpenAI models, while the Treasury Department and the Department of Health and Human Services initiated plans to phase out Claude.

Anthropic filed suit, asserting that the administration retaliated against the company for its safety positions and infringed on its due process protections. Government attorneys argued that vendor-mandated usage restrictions introduce operational instability, creating uncertainty as to whether essential AI software would remain accessible during military engagements.

In August, U.S. District Judge Rita Lin permanently enjoined the Pentagon's designation, ruling that invoking national security did not grant officials authority to punish private contractors over policy stances. The appellate court's Friday decision overturns that halt, placing the supply-chain restriction back into effect.

The litigation highlights the broader friction between federal procurement demands and proprietary AI governance. Competitor OpenAI previously secured a classified network deployment agreement containing its own terms against domestic surveillance and autonomous targeting, while publicly opposing Anthropic's blacklisting.

Although White House and procurement officials have held ongoing discussions with Anthropic as the startup explores a potential initial public offering, the appellate ruling leaves substantial government contract revenue at risk and sets an influential precedent for commercial software terms in federal contracting.