San Francisco judge pauses Pentagon’s supply-chain risk label for Anthropic, blocking broader federal-use restrictions temporarily

A court order temporarily halts a rare national-security procurement tool

A federal judge in San Francisco has temporarily blocked the U.S. Department of Defense from enforcing a “supply-chain risk” designation against Anthropic, a major artificial intelligence company headquartered in San Francisco. The ruling also pauses enforcement of a separate directive tied to the same dispute that sought to halt federal agencies’ use of Anthropic’s technology, including its Claude chatbot, while litigation continues.

The order was issued by U.S. District Judge Rita Lin following a hearing earlier in the week. In written findings, the court signaled skepticism that the government’s measures were aligned with the stated national-security purpose of the underlying procurement authority, emphasizing that the dispute is being evaluated through the lens of administrative law and constitutional limits on government action.

How the conflict escalated from contracting terms to a government-wide stigma

The case centers on a breakdown in negotiations over how Anthropic’s AI models could be used in defense contexts. Anthropic has sought to limit certain uses of its technology, including fully autonomous weapons and forms of surveillance involving Americans. The Defense Department has taken the position that it should be able to employ the tools it procures in any manner it deems lawful.

In late February 2026, federal actions escalated rapidly. The administration publicly criticized the company and moved to restrict federal use of Anthropic’s products. The Defense Department then invoked a statutory mechanism that can exclude a vendor from defense procurements involving national security systems to reduce supply-chain risk. Anthropic argues the designation functioned as a punitive measure that threatened its reputation, commercial relationships, and ability to work with federal contractors.

What “supply-chain risk” can mean for contractors and partners

In practical terms, a supply-chain risk label can ripple beyond a single government contract. It may require contractors, suppliers, or partners supporting the U.S. military to certify they are not using the designated vendor’s covered products or services in relevant systems, prompting re-evaluations of toolchains and subcontracts.

• Defense contractors may pause or unwind deployments to avoid compliance exposure.
• Vendors embedded in larger systems can face downstream de-risking by partners.
• Even temporary uncertainty can trigger procurement delays and contract amendments.

Key points in the judge’s reasoning and what happens next

The court characterized the challenged measures as unusually broad and potentially debilitating for the company. The judge’s order provides temporary relief while the merits are litigated and does not compel the Defense Department to use Anthropic’s products. The ruling also includes a short delay before taking effect, giving the government time to consider next steps.

The legal fight now turns to whether the Defense Department’s designation fits the statutory criteria for supply-chain risk actions, and whether required procedures were followed.

Anthropic is pursuing parallel litigation tracks, including a separate proceeding in a federal appeals venue focused on a different legal pathway the government has cited for the designation. The Defense Department’s response in court will shape whether the temporary pause becomes a longer-lasting injunction or is narrowed or reversed as the case progresses.

For San Francisco’s AI sector and federal procurement more broadly, the dispute is likely to clarify how far national-security procurement authorities can extend when a vendor challenges proposed terms for the military use of rapidly evolving AI systems.