Anthropic Supply Chain Risk Ruling Explained Simply

On September 25, 2026, a federal appeals court in Washington, D.C. ruled against Anthropic in its fight with the Pentagon. The court upheld the government’s decision to label the AI company a “supply chain risk,” which stops the US military and its contractors from using Claude in defense work.

If you use Claude, work for a company that does government contracts, or just want to understand why an American AI company is fighting its own government, this guide explains what happened in plain words.

Quick Answer

• A US appeals court ruled 2 to 1 on September 25, 2026 that the Pentagon can keep Anthropic on its supply chain risk list
• The label bars the military and defense contractors from using Claude in their defense work
• Anthropic says the ruling does not affect regular users or commercial customers
• A separate federal judge in San Francisco struck down a different part of the government’s action in August, and that ruling still stands
• Anthropic says it disagrees and is considering all options, including asking for another review
• The court delayed the ruling from taking effect so Anthropic has time to appeal

What Is a “Supply Chain Risk”

A supply chain risk label is a government tool meant to protect its systems from vendors that could threaten national security. When the Pentagon uses it on a company, the military and the companies that work for it are blocked from buying or using that company’s products in defense projects.

It is normally used for foreign companies suspected of spying or sabotage. Using it on a major American AI company is unusual, which is why this case has drawn so much attention.

How Did We Get Here

• July 2025: Anthropic signed a $200 million contract with the Pentagon. Claude became one of the first advanced AI models approved for use on classified government networks.
• Late 2025: Talks over expanding that work broke down. The Pentagon wanted the right to use Claude for any lawful purpose. Anthropic wanted two exceptions written in: no mass domestic surveillance of Americans and no fully autonomous weapons.
• February 27, 2026: President Trump told federal agencies to stop using Anthropic’s technology, and Defense Secretary Pete Hegseth announced the supply chain risk label. A formal letter followed in early March.
• March 2026: Anthropic sued. Because the government used two separate legal authorities, the case was split into two courts: one in San Francisco and one in the D.C. appeals court.
• August 2026: A San Francisco judge, Rita Lin, ruled against the government on one of the two designations. She found the government had not shown what that law requires. That ruling remains in place.
• September 25, 2026: The D.C. appeals court upheld the other designation.

What the Appeals Court Decided

The three-judge panel voted 2 to 1. Judge Gregory Katsas wrote the majority opinion, joined by Judge Neomi Rao. Judge Karen Henderson disagreed.

The majority said the Pentagon had enough support for its conclusion that using Claude in its systems, directly or through contractors, was a national security risk covered by the law. It accepted that the Pentagon can consider whether restrictions built into a vendor’s AI could get in the way of military operations. It also said the government does not need to prove bad motive on Anthropic’s part to use this law.

The court rejected Anthropic’s arguments that the ban was arbitrary, unauthorized, and unconstitutional. It also delayed the ruling from taking effect right away, so Anthropic can ask the same panel to reconsider, or ask the full D.C. court to hear the case.

The Two Sides of the Argument

The government’s position: The military needs to decide how its tools are used, and a vendor cannot place limits on lawful military operations. The court agreed that the Pentagon could treat those restrictions as a risk to its systems.

Anthropic’s position: The company says it was punished for refusing to allow its AI to be used for lethal autonomous warfare or mass surveillance of Americans. It argues the government violated its First Amendment rights, misused a national security law, and skipped the normal process for ending federal contracts. In the San Francisco case, Judge Lin found that the government’s actions appeared aimed at making a public example of Anthropic for criticizing it, rather than being based on real evidence that it would sabotage its own model.

Both sides describe the fight as a matter of principle, and the courts have so far split, one on each designation.

Does This Affect You as a Claude User

According to Anthropic’s own statements since the label was first announced, the answer is no for most people. The company says the designation only covers the use of Claude as a direct part of Department of War contracts. Individual users and companies with normal commercial contracts keep full access to Claude through its apps and API.

If your company does defense work: The label can apply to Claude used inside covered defense contracts. Law firms have advised contractors to review where Claude or other Anthropic products are used in their government work, and to ask their own lawyers what applies to them. This is general information, not legal advice.

If you are a regular user: Nothing changes today. You can still use Claude the same way. For what is currently available and what it costs, see our Claude Opus 5.5 and GPT-6 comparison.

What Happens Next

• Anthropic can ask the same panel to reconsider, or ask all the judges of the D.C. court to hear the case
• It could also ask the US Supreme Court to take it up
• The San Francisco ruling in Anthropic’s favor on the other designation remains in effect
• A news brief on September 28 reported that President Trump said he plans to meet Anthropic CEO Dario Amodei

The two designations are treated separately by different courts, so the overall picture is mixed. The government won this round, and Anthropic won the earlier one.

The Bigger Picture

This case is part of a tense relationship between Anthropic and the current administration. Amodei recently drew criticism from President Trump for calling for an industry-wide slowdown, an idea we explained in our guide to his Pace the Frontier plan. At the same time, Anthropic’s tools are widely used by businesses, including large banks, as we covered in our piece on the JPMorgan Claude spending cap.

Frequently Asked Questions