A federal judge in San Francisco has temporarily blocked the Pentagon from branding AI company Anthropic a national security threat, ruling that the designation was likely unlawful and may have been designed to punish the company for speaking out against military uses of its technology.
U.S. District Judge Rita Lin granted a preliminary injunction pausing the Department of War's decision to label Anthropic a "supply chain risk," a designation that threatened to cut the company off from federal contracts and ripple through its commercial partnerships. The ruling hands an early courtroom loss to Defense Secretary Pete Hegseth, who has already faced judicial pushback on other enforcement actions he has pursued since taking office.
The clash traces back to a contract fight over how the military could deploy Anthropic's Claude AI system. Anthropic had placed limits on Claude's use, requesting that it not be deployed for mass domestic surveillance or fully autonomous weapons. The Pentagon viewed those restrictions as unacceptable for a defense contractor. When negotiations broke down, the administration labeled Anthropic a supply chain risk, a classification that could block the company from up to $200 million in existing contract work and bar other agencies from using Claude.
Anthropic sued, alleging that Hegseth exceeded his authority and that the designation amounted to government retaliation against the company's public stance on AI safety. The company claimed the label was already causing immediate business and reputational harm, pushing partners to reconsider contracts and leading agencies to stop using Claude, as Breitbart reported.
Judge Lin did not mince words from the bench. She said the Pentagon's designation looked like an attempt to "cripple" Anthropic and possibly punish it for seeking public scrutiny of the contracting dispute.
"It looks like DoW is punishing Anthropic for trying to bring public scrutiny to this contracting dispute, which, of course, would be a violation of the First Amendment."
In her written order, Lin went further. She rejected the government's core legal theory for the designation, calling it fundamentally unsupported by the statute that authorizes supply chain risk labels.
"Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government."
That language, drawn from CBS News reporting on the ruling, signals that the judge views Anthropic's First Amendment claim as strong enough to survive early litigation. Lin found the designation was likely "arbitrary and capricious," the legal standard courts use when an agency acts without a rational basis or ignores its own rules.
One of the most damaging moments for the administration came not from the judge but from its own attorney. During the hearing, Trump administration lawyer Eric Hamilton acknowledged a gap in the government's legal position that cut against Hegseth's public statements about the scope of the ban.
"Standing here today, I'm not aware of any authorities that would permit DoW to categorically bar contractors from using a company's products or services for non-DoW work."
That admission matters because Hegseth had signaled publicly that military contractors could not do business with Anthropic at all, not just on Pentagon projects. The Washington Examiner reported that Hamilton's concession undercut the broader sweep of the secretary's position and gave the judge additional reason to question whether the designation was tailored to a legitimate security concern or designed as a blunt instrument against a company that refused to comply.
Anthropic's lawsuit rests on two pillars. First, the company argues the supply chain risk designation violated federal procurement law because it was imposed without the procedural safeguards and factual basis the statute requires. Second, Anthropic claims the government violated its First Amendment rights by retaliating against the company's public advocacy for AI safety guardrails.
The First Amendment angle is the more unusual of the two. Corporate speech claims against the federal government in the procurement context are rare, and courts have generally given agencies wide latitude in deciding which companies pose security risks. But Anthropic's lawyers framed the case as something different: not a routine contract dispute, but a case where the government weaponized a security label to silence a company that disagreed with its policies.
An Anthropic spokesperson welcomed the ruling.
"We're grateful to the court for moving swiftly, and pleased they agree Anthropic is likely to succeed on the merits."
The preliminary injunction means the supply chain risk label is frozen while the case proceeds. Anthropic can continue operating under its existing contracts, and agencies are not required to stop using Claude based on the now-paused designation.
The Pentagon did not accept the ruling quietly. Under Secretary of War Emil Michael posted on X that the decision "contained dozens of factual errors" and "seeks to upend the (president's) role as Commander in Chief," Fox News reported. That framing sets up a likely appeal and positions the administration to argue that the judiciary is overstepping into national security decisions that belong to the executive branch.
The government's defense throughout the hearing centered on contract and security concerns rather than retaliation. Pentagon officials maintained that the supply chain risk designation was a legitimate response to a contractor that refused to meet the military's operational needs. Whether that argument holds up as the case moves forward will depend on the factual record the government can produce to justify the label.
The preliminary injunction is not a final ruling. It means only that Judge Lin found Anthropic is likely to succeed on the merits and that the company would suffer irreparable harm without court intervention. The case will now proceed through discovery and potentially a full trial, where the government will need to produce internal communications and decision-making records that explain how and why the designation was imposed.
Investigators and litigators on both sides will need to determine several key questions. Authorities have not publicly confirmed whether internal Pentagon communications show the designation was tied to Anthropic's public statements on AI safety or to genuine security assessments. The administration has not disclosed the specific factual basis it relied on when classifying Anthropic as a supply chain risk. And no public results have been released about whether other AI companies that imposed similar use restrictions faced comparable treatment.
The case also raises broader questions about how far any administration can go in punishing companies that set ethical boundaries on their technology. If the government can strip a company of its contracts and blacklist it from federal work simply because it refused to allow certain military applications, the chilling effect on corporate speech could extend well beyond the AI sector.
This is not the first time a federal judge has stepped in to check executive action during the current administration. Courts have been active in reviewing a range of government decisions, from grand jury procedural rewrites in Washington to challenges against agency enforcement. For Hegseth personally, the Anthropic ruling adds to a growing list of courtroom setbacks that test the boundaries of his authority at the Pentagon.
The New York Post noted that the temporary block keeps the status quo in place while litigation unfolds, but the real fight is just beginning. If the case reaches the appellate level, it could produce binding precedent on whether supply chain risk designations are subject to meaningful judicial review or whether courts must defer to the Pentagon's judgment on which companies threaten national security.
When the government can destroy a company's business with a single label and no meaningful process, the question stops being about AI policy. It becomes about whether any American business can speak freely without fear of federal retaliation.
