Judge rules the Pentagon’s supply chain risk label for Anthropic unlawful

Federal Court Strikes Down Pentagon’s Supply Chain Risk Designation Against Anthropic

Earthguardiansonline.com – A Northern District of California judge ordered the Department of Defense to remove its “supply chain risk” label from AI developer Anthropic on Thursday evening, declaring the designation unlawful. The ruling, issued by Judge Rita Lin, a Biden-era appointee, effectively ends a months-long standoff between the world’s largest military spender and one of the most prominent artificial-intelligence firms in the country. The decision lands at a moment when Washington is racing to integrate machine-learning tools into everything from battlefield decision-making to signals intelligence, and it signals that the executive branch cannot wield procurement designations as a blunt instrument of political retribution.

What a “Supply Chain Risk” Label Actually Means

In U.S. defense procurement, a supply chain risk designation functions as a near-total exclusion order. Once applied, no component of the Pentagon — whether a combatant command, a logistics office, or a private contractor holding a DoD contract — may purchase, license, or integrate products from the named firm. Historically, the label has been reserved for companies perceived to have ties to foreign adversaries: firms with ownership stakes in China, Russia, or other nations the United States regards as strategic competitors. Applying it to a domestic startup that had never been accused of foreign entanglement was, by the department’s own precedent, unprecedented.

How the Dispute Began

The confrontation traces back to early this year, when Anthropic declined to strip away internal safety guardrails embedded in its Claude language model. Those guardrails were designed to prevent the model from being deployed in fully autonomous weapons systems or in mass-surveillance operations. Anthropic’s position was straightforward: its models had not yet reached a reliability threshold sufficient for tasks where errors could cost lives or erode civil liberties. The company was, in effect, telling the Pentagon that its software was not ready for the most consequential military applications.

Defense Secretary Pete Hegseth pushed back publicly, arguing that the American military could not be told by a private corporation how to wield its own tools. In February, his department took the extraordinary step of slapping the supply chain risk label onto Anthropic, cutting the company off from every Pentagon procurement channel overnight.

The Court’s Reasoning

Anthropic filed suit in March. Judge Lin, writing for the Northern District of California, rejected the Pentagon’s central justification — that the department simply could not “trust” Anthropic with sensitive work. In her opinion, Lin made clear that a bare assertion of national-security concern does not insulate an agency from constitutional scrutiny.

“The empty invocation of national security is not a blank check to punish and retaliate against government critics.”

She went further, finding that the documentary record indicated the department’s true motive was to make a public example of Anthropic for what she characterized as its “arrogance” in criticizing government policy, rather than any articulable basis to believe the company would actually sabotage its own model.

“[Anthropic] constituted unlawful retaliation in violation of the First Amendment” and “was denied the pre-deprivation process required under the Fifth Amendment.”

The First Amendment prong of the ruling is particularly pointed. It holds that a company which publicly criticizes a government agency cannot then be economically punished by that same agency through procurement exclusion, absent a genuine, articulable security rationale. The Fifth Amendment prong addresses procedure: before the government strips a firm of its ability to earn revenue from federal contracts, due process requires notice and an opportunity to be heard. Anthropic received neither.

Other Agencies Kept Working With Anthropic

Lin noted a telling inconsistency in the government’s own conduct. Even after the Pentagon imposed the label, other federal departments continued to meet with, contract with, and deploy Anthropic’s technology. That parallel activity undercut any claim of a bona fide, government-wide fear that the company was a saboteur.

“None of that is consistent with a genuine fear that Anthropic is a saboteur who would poison its software to harm national security.”

Political Backdrop and Aftermath

The dispute had drawn presidential attention. In a June interview with Axios, President Donald Trump acknowledged that he had once regarded Anthropic as a national-security threat but said he no longer held that view. The comment, while not legally binding, added political pressure on the department to reconsider its posture.

An Anthropic spokesperson issued a brief statement welcoming the decision.

“We remain focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology.”

The Pentagon had not responded to a request for comment as of Thursday evening.

What Remains Unresolved

The ruling does not close every legal front. A second lawsuit brought by Anthropic, also connected to the supply chain risk label, is still pending before a court in Washington, D.C. That case may address additional procedural and substantive questions about how long the designation remained in force and what damages or injunctive relief follow from its duration.

For the broader AI industry, the decision establishes a cautionary precedent: the supply chain risk tool, however useful for genuine foreign-ownership concerns, cannot be repurposed as a retaliatory sanction against domestic firms that push back on how their technology is deployed. As defense budgets grow ever more dependent on commercial machine-learning products, the line between legitimate procurement judgment and unconstitutional punishment will be tested repeatedly. Thursday’s ruling draws that line, at least for now, in Anthropic’s favor.