The Pentagon’s effort to force broader military access to Anthropic’s AI has run into a constitutional wall. A federal judge found the government crossed it when it branded the company a national-security and supply-chain threat.

The clash began over the winter, when Defense Secretary Pete Hegseth sought revised contracts allowing military use of AI for “any lawful use.” Most major labs accepted the terms, but Anthropic refused to permit mass surveillance of Americans or fully autonomous weapons, arguing those applications posed a narrow but serious democratic risk.¹

The standoff escalated into public pressure, an ultimatum and, ultimately, the Pentagon’s designation of Anthropic as a “supply chain risk” — a label that threatened to shut the firm out of government business and dealings with military contractors. Anthropic sued in March, contending that the designation was retaliation for its public opposition to the Pentagon’s demands.²

On Thursday, U.S. District Judge Rita Lin struck down the blacklist. “The empty invocation of national security is not a blank check to punish and retaliate against government critics,” she wrote, finding the designation arbitrary, capricious and a violation of the First Amendment.¹ Lin also noted the Pentagon continued pursuing work with Anthropic after calling it a threat — conduct she said did not square with a genuine fear that the company would sabotage its software.²

Anthropic welcomed the ruling and said it remained focused on working with government on national security. The administration is expected to appeal, and a separate challenge to another Pentagon designation remains before the D.C. Circuit.² The immediate case is a clear win for Anthropic; the wider fight is over who gets to set the limits on military AI.