Judge blocks Pentagon penalties against Anthropic as illegal

U.S. District Judge Rita Lin issued a 59-page written order Thursday night blocking the Pentagon’s measures against Anthropic after finding the government acted illegally in punishing the AI company for criticizing Defense Department AI policy. The ruling freezes enforcement tied to a directive telling federal agencies to stop using Anthropic and its chatbot Claude, with the administration expected to contest the decision.

Key Takeaways

A 59-page written order by U.S. District Judge Rita Lin found the Pentagon acted illegally in penalizing Anthropic after branding it a “supply chain risk,” framing the move as retaliation for the company’s public criticism of the administration’s AI views.
The court blocked the government’s measures, including enforcement tied to a directive ordering federal agencies to stop using Anthropic and its chatbot Claude.
The dispute traces to February 2026 accusations by President Donald Trump and Defense Secretary Pete Hegseth that Anthropic endangered national security, after CEO Dario Amodei raised concerns about mass surveillance and autonomous armed drones.
A separate, narrower Anthropic case remains pending at the federal appeals court in Washington, D.C., involving a different Pentagon rule used to try to declare the company a supply chain risk.

Judge Lin’s 59-Page Order Blocks Pentagon’s Anthropic Penalties

U.S. District Judge Rita Lin issued a written order Thursday night siding with Anthropic and blocking the Pentagon’s measures against the company after the government labeled it a supply chain risk earlier this year. Lin concluded the Pentagon acted illegally by punishing Anthropic for criticizing the Department of Defense’s views on AI use, a framing that turns a procurement dispute into a constitutional constraint on how national-security tools can be deployed.

The injunction matters operationally because it blocks enforcement tied to a directive ordering federal agencies to stop using Anthropic and its chatbot Claude. The White House did not immediately respond to a request for comment as of publication.

Anthropic said it welcomed the ruling. “We remain focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology,” an Anthropic spokesperson said.

How the ‘Supply Chain Risk’ Label Became a First Amendment Fight

The mechanism at the center of the case is the “supply chain risk” designation, a government label that can restrict or discourage federal agencies from buying or using a vendor’s products due to alleged security concerns. In February 2026, President Donald Trump and Defense Secretary Pete Hegseth accused Anthropic of endangering national security and designated the company a supply chain risk.

Anthropic CEO Dario Amodei refused to back down, citing concerns the company’s products could be used for mass surveillance or autonomous armed drones. That public posture became the factual spine of Anthropic’s retaliation theory: the company argued the government’s procurement and security apparatus was being used to punish speech and to pressure Anthropic to allow broader military use of its technology.

Anthropic sued the Pentagon in March, calling the designation and related actions an “unlawful campaign of retaliation” tied to its refusal to allow unrestricted military use. Lin had already temporarily blocked the Pentagon from labeling Anthropic a supply chain risk and blocked enforcement of Trump’s social media directive ordering all federal agencies to stop using Anthropic and Claude while the case proceeded.

Inside the Court’s Reasoning: ‘No Articulable Basis’ for Sabotage Claims

Lin’s written order goes beyond a procedural pause and attacks the government’s justification on the record in front of her. In the 59-page ruling, she wrote that neither the Constitution nor the federal statute the government invoked allows it to “impose sweeping penalties based principally on Anthropic’s critique of the Administration’s views.”

The sharpest line in the order is about motive and evidentiary support. Lin wrote the government’s actions “were based on a desire to make a public example out of Anthropic for its 'arrogance' in criticizing the government, not based on any articulable basis to believe that Anthropic would actually sabotage its model.” That finding undercuts the credibility of the “supply chain risk” rationale in this specific dispute, because it treats the national-security framing as pretext rather than substantiated risk.

The government previewed part of its defense at a July 30 hearing, where Lin said the government’s position was “really troubling” and seemed “at odds to me with the First Amendment.” She also said the record had “gotten worse for the government” over time. Department of Justice lawyers argued that AI models are “so staggeringly enormous and opaque” that the Defense Department cannot evaluate them the same way it would a physical piece of hardware, while Anthropic lawyer Michael Mongan said the actions “profoundly harm Anthropic” and “threaten more broadly to chill speech and debate on a very important issue.”

Appeal Risk, Parallel Litigation, and the OpenAI Pentagon Deal Backdrop

The immediate uncertainty now is procedural: the government is expected to fight Lin’s decision, but the public record here does not specify whether an appeal has been filed or what the next deadlines are. That leaves procurement teams and vendors in a familiar limbo where the injunction is real, but the durability of the ruling depends on what happens next in appellate posture.

There is also a second track. Anthropic has filed a separate and narrower case that remains pending in the federal appeals court in Washington, D.C. That matter involves a different Pentagon rule being used to try to declare Anthropic a supply chain risk, which means the government still has another legal pathway to test similar theories even if Lin’s order holds.

The competitive backdrop is unusually visible for a procurement fight. Anthropic’s primary tech industry rival OpenAI made its own deal to work with the Pentagon just hours after the government punished Anthropic for its stance, and both companies are described as ramping up for initial public offerings. The terms of OpenAI’s Pentagon deal are not detailed in the record provided here, but the sequencing makes access to government contracts and reputational signaling part of the market narrative around “defense AI.”

Forward signals now cluster around four items: whether DOJ formally appeals Lin’s written order and on what timeline, how the D.C. appeals-court matter progresses, whether the Defense Department revises guidance for evaluating AI models as “supply chain risk” in light of DOJ’s opacity argument, and whether further disclosures emerge about Pentagon partnerships with major AI labs as IPO preparations accelerate.

My Read: Procurement Retaliation Claims Are Now a Live Risk Factor for ‘Defense AI’ Narratives

The threshold that matters is whether this order survives contact with an appeal, because Lin did not just disagree with the Pentagon’s risk assessment. She treated the procurement tool as a speech penalty, and she did it with language that says the government lacked an “articulable basis” for the sabotage theory and instead wanted a public example.

If DOJ fights and loses, the setup starts to look structural rather than narrative-driven: agencies will still have national-security levers, but they will have to build a record that looks like security analysis rather than message discipline, especially when vendors are publicly debating surveillance and weapons use.

Judge blocks Pentagon penalties against Anthropic as illegal retaliation