A federal appeals court on Friday upheld the Pentagon’s decision to blacklist Anthropic from military contracts, delivering a victory to President Donald Trump and War Secretary Pete Hegseth.
The dispute centers on safety guardrails the AI startup placed on how its technology may be used.
The U.S. Court of Appeals in Washington ruled 2 to 1 against Anthropic in the company’s challenge to its March designation as a national security supply chain risk.
Anthropic says the designation has cost it billions of dollars in lost business and damaged its reputation before a highly anticipated initial public offering.
The court majority concluded that the Pentagon acted reasonably after Anthropic refused to permit use of its products for autonomous weapons or mass surveillance.
The judges also rejected the company’s claim that the Pentagon retaliated against it for its positions on AI safety and ethics.
Hegseth argued that the startup’s safety restrictions could compromise military operations, and the majority sided with that concern.
Judge Gregory Katsas wrote, “The Secretary raises the deeply sobering prospect of overly constrained AI models shutting down unexpectedly and thus causing important military operations to fail,” in the opinion.
Anthropic, which is based in San Francisco, said Friday that it respectfully disagrees with the decision.
The company said it remains confident in its position and is considering its options, including asking the full appeals court to review the decision issued by the panel of three judges.
The White House and the Department of War did not immediately respond to emails seeking comment after the ruling, according to the source report.
The appellate result follows a separate ruling last month by a federal judge in San Francisco involving a parallel designation under another law.
In that case, U.S. District Judge Rita Lin found that the administration had unlawfully retaliated against Anthropic over its views on AI safety.
Lin also blocked Trump’s ban on Anthropic across the government. Her order further blocked Hegseth’s directive that barred military contractors from conducting any business with the startup.
Hegseth designated Anthropic as a risk under two different laws after the company declined to remove its guardrails.
Those limits were intended to prevent the technology from being used for autonomous weapons or for mass domestic surveillance.
The designations followed months of negotiations that ended without an agreement and gave way to public recriminations from Hegseth and Trump.
Both accused Anthropic of endangering American lives through ideological posturing, a characterization that the company denied.
Anthropic maintained that AI technology is not yet reliable enough to be used safely in autonomous weapons.
