Court Says Pentagon Broke the Law by Blacklisting AI Company Anthropic

A federal judge has ruled that the Pentagon broke the law when it blacklisted AI company Anthropic from government contracts, calling the move an unconstitutional punishment for the company's free speech.

Judge Rita F. Lin, who serves on the U.S. District Court for the Northern District of California, issued her ruling on August 28, 2026. The decision settles the main claims in a lawsuit Anthropic filed in March, Anthropic PBC v. U.S. Department of War et al., No. 3:2026cv01996, which examined how far the government can go when it buys technology from private companies (The Verge).

The conflict started in fall 2025. The Department of Defense wanted what it called "all lawful uses" of Anthropic's AI system, called Claude. Anthropic said yes to a military contract but refused to remove two rules: the company would not allow its AI to be used for mass surveillance of Americans, and it would not allow its AI to be used in lethal autonomous weapons, meaning weapons that can select and attack targets on their own (Washington Technology).

Defense Secretary Pete Hegseth responded by ordering all AI companies with military contracts to allow "any lawful use" of their technology. He then designated Anthropic a supply chain risk. This label is a formal classification that effectively bans a company from doing business with the federal government. After blacklisting Anthropic, the Pentagon signed deals with seven other AI labs, including Google, Microsoft, OpenAI, and SpaceX, to take over the role Anthropic had been filling (The Verge).

Anthropic filed its lawsuit in March, arguing that the Trump administration had punished the company for setting limits on how its AI could be used. The company did not ask the court to force the government to accept its contract terms. Instead, it asked the court to strike down the policies that had blacklisted and excluded it from government contracting (The Verge; Justia).

In March, Judge Lin granted a temporary order blocking the blacklist. She wrote that the Department of War designated Anthropic a supply chain risk because of its "hostile manner through the press," and that punishing a company for bringing public scrutiny was "classic illegal First Amendment retaliation" (The Verge). Her final ruling confirms that earlier assessment.

The government fought the retaliation claim throughout the case. In a March 17 court filing, the Trump administration denied that the blacklisting was unlawful retaliation, arguing it was justified and lawful (Reuters).

The case also went through a separate appeals court. On April 8, 2026, the U.S. Court of Appeals for the D.C. Circuit declined to block the Pentagon's blacklisting and set oral arguments for May 19, 2026, directing the parties to address three specific questions. Anthropic had asked that court to review the Pentagon's designation, arguing it was unconstitutional retaliation (Reuters; CNBC). The California case continued on a separate track, focused on striking down the blacklist policies themselves rather than the contract dispute before the D.C. Circuit (Jones Walker; Justia).

Judge Lin's final ruling found that Hegseth's designation was arbitrary and capricious, a legal standard meaning the government failed to provide a sensible, reasoned basis for its decision. The First Amendment retaliation finding addresses the constitutional problem of punishing a company for its speech. The arbitrary and capricious finding addresses the procedural problem of the government acting without a rational justification (The Verge).

The broader context here is that the Department of Defense has been pushing aggressively to get unrestricted access to AI from multiple companies. The fact that the Pentagon signed deals with seven replacement labs after blacklisting Anthropic shows the government had other options, which weakens any argument that excluding Anthropic was necessary rather than punitive.

What this ruling does is establish that AI companies keep their First Amendment rights when they publicly state and defend limits on how their technology is used, even when the government says national security is at stake. For companies that have earned customer trust by promising safety safeguards, the decision provides a constitutional backstop against government agencies that might try to use their purchasing power to override those promises.

Whether the government appeals Judge Lin's ruling, and how the separate D.C. Circuit case interacts with this one, will shape what happens next. The tension between government demand for unrestricted AI and companies' own usage limits will not be settled by a single court decision. But Judge Lin's ruling establishes that the Constitution does not simply side with the government because national security procurement is involved.