A divided U.S. appeals court ruled on September 25 that the Trump administration had the authority to blacklist Anthropic under a broad provision of the Supply Chain Security Act, upholding a Pentagon action that banned federal agencies and contractors from using the company’s AI technology. The 2-1 decision by the U.S. Court of Appeals for the D.C. Circuit deepens a legal conflict over the government’s ability to restrict suppliers that refuse to enable military applications.
The ruling conflicts with a Northern District of California decision last month that found the designation unlawful under a different statute. Judge Rita Lin determined the blacklisting violated the First Amendment and retaliated against Anthropic for refusing to remove restrictions on lethal autonomous warfare and mass surveillance, Ars Technica reported. The appeals court acknowledged the California court’s findings but said its own review was limited to a separate, more permissive grant of authority.
The legal fight began in March when the Pentagon ordered a halt to federal use of Anthropic’s products and blocked defense contractors from doing business with the company. While the courts examined the blacklisting, Anthropic continued to release new models: as Automatica reported, Claude Opus 5.5 debuted on Amazon Bedrock in late September.
The D.C. Circuit majority, written by Trump appointees Gregory Katsas and Neomi Rao, said the blacklisting was permitted under 41 U.S.C. § 4713, which defines a “supply chain risk” to include any person who may “deny” the function of covered technology. Unlike the law the California court reviewed, this provision does not require a finding of malicious intent by an adversary. “Anthropic disabling Claude from performing lawful actions requested by the Department qualifies as a ‘supply chain risk,’” the majority held, according to the ruling.
In a dissent, Judge Karen Henderson, a George H.W. Bush appointee, argued the law should be read in context to cover deliberate sabotage or spying, not a contractor’s upfront restrictions. “The verbs at issue are all directed at deliberately impeding or eavesdropping on the ‘function, use, or operation’ of a covered article,” she wrote.
Anthropic said it “respectfully disagree[s] with the court’s decision” and is considering all options, including a full-court review or a Supreme Court petition. Commerce Secretary Howard Lutnick recently told reporters that the administration and Anthropic have “patched up” their relationship, though the blacklisting remains in place.
The Northern District of California ruling by Judge Lin, a Biden appointee, had concluded the blacklisting violated Section 3252, which is limited to malicious acts by adversaries. Lin also found the action was unconstitutional retaliation. “The empty invocation of national security is not a blank check to punish and retaliate against government critics,” she wrote.