U.S. District Judge Rita Lin indicated at a July 30, 2026 hearing that the Trump administration still lacks sufficient evidence to justify its supply chain risk designation against Anthropic, keeping alive the AI company's challenge to a Pentagon ban that has barred federal agencies from using its technology.

The hearing in San Francisco marks the latest chapter in a dispute that began when Secretary of Defense Pete Hegseth declared Anthropic a supply chain risk, the first time a U.S. company was publicly designated under the relevant government authority Reuters. The Defense Department confirmed the designation in a letter to Anthropic on March 4, 2026 Anthropic. The label effectively banned the federal government from using Anthropic's AI products.

The dispute traces back to stalled contract negotiations between Anthropic and the DOD. Anthropic CEO Dario Amodei stated publicly that the Pentagon threatened to remove the company from federal systems if it maintained certain AI safeguards Anthropic. Specifically, Anthropic refused to allow its models to be used for mass surveillance of Americans or for lethal targeting and firing decisions TechCrunch.

Anthropic filed two lawsuits against the DOD in March 2026: one challenging the ban and the supply chain risk designation, and a second seeking review of the designation under a separate statute, according to a Congressional Research Service report Congress.gov. That second suit is being heard in Washington.

Judge Lin issued a ruling on March 26, 2026, temporarily blocking the Pentagon's supply chain risk designation and prohibiting federal agencies from applying the label Reuters. Senator Mark Warner welcomed the decision, issuing a press release praising the court's pause on Hegseth's designation Senate.gov. The Pentagon has since appealed Lin's March 26 order Inside Defense.

At the D.C. Circuit Court of Appeals, the bench appeared split during oral arguments held on or around May 19, 2026 AP News. The appeals court generally seemed skeptical of Anthropic's bid to maintain the block, though Judge Karen LeCraft Henderson stated she saw no evidence supporting the Pentagon's determination that Anthropic poses a supply-chain risk Bloomberg; AP News.

At the July 30 hearing, Judge Lin weighed whether to make her temporary block permanent. The government advanced two principal arguments for the designation. First, the DOD argued that Anthropic's public criticism of the Pentagon justified the supply chain risk ban. Lin called that argument "really troubling" and warned it could set a precedent of retaliating against federal contractors who disagree with the administration TechCrunch.

Second, the DOD claimed Anthropic could potentially disable or alter its AI models during warfighting operations, implying an operational risk if the models were deployed in combat scenarios. Lin stated she saw no proof that Anthropic could alter a delivered model or activate a kill switch TechCrunch.

While the legal battle has played out across two federal courts, the White House has been working on a parallel track. The administration is developing guidance that would allow federal agencies to circumvent the supply chain risk designation and onboard new uses of Anthropic's technology Nextgov; Axios. Concurrently, the Trump administration is drafting an AI executive order touching on related federal AI procurement issues.

The Congressional Research Service published its own analysis of the legal disputes in May 2026, cataloging the dual-lawsuit strategy and the statutory questions each case raises Congress.gov.

The central tension here is not subtle. A defense secretary used a supply chain risk authority, originally designed to protect federal procurement from foreign-controlled or compromised suppliers, against a domestic AI company whose stated objection was to limits on military use of its own products. The DOD's argument that public criticism of the Pentagon could itself constitute a supply chain risk, if accepted, would extend the designation's reach into contractor speech in a way no prior case appears to have tested. And the kill-switch claim, which Lin found unsupported by evidence, raises a genuine technical question worth attention: whether a model provider retains post-deployment control over a model running inside a government environment depends entirely on the deployment architecture, and the facts before the court did not establish that Anthropic had such capability.

What this case ultimately turns on may be less about Anthropic's specific safeguards and more about whether the supply chain risk framework can be applied as leverage in a contract dispute over acceptable use terms. If Judge Lin makes her injunction permanent, the Pentagon's appeal will continue at the D.C. Circuit, where the panel's skepticism in May cut against Anthropic but did not clearly favor the government either. The split bench suggests the appellate outcome is genuinely uncertain.