US Appeals Court Upholds Pentagon's Ban on Anthropic

A US appeals court upheld on Friday the Pentagon's decision to exclude Anthropic from its supply chain, a decision taken after the AI startup refused to remove certain safeguards on its Claude model. The dispute began in February when Anthropic refused Defense Secretary Pete Hegseth's demand to lift restrictions on using Claude for fully autonomous lethal weapons and mass domestic surveillance of Americans. The formal supply chain risk designation, made in March, cancelled Anthropic's military contracts and barred other Pentagon contractors from using its technology.

What Triggered the Alarm

According to the court's decision, an Anthropic executive had questioned Pentagon contractor Palantir's use of Claude during the military operation that captured Venezuelan President Nicolas Maduro on January 3. Under Secretary of Defense Emil Michael said this objection raised doubts about whether Anthropic might disable its software mid-operation, endangering service members.

The Court's Reasoning

Anthropic maintains it cannot modify its models once delivered to the military, but the appeals court noted the company controls each new version's behavior, and the Pentagon must stay current with the latest offerings. Writing for two of three judges, Judge Gregory Katsas said it falls to the president and defense secretary to weigh such risks.

A Conflicting Ruling Elsewhere

This decision conflicts with a separate California federal court ruling last month that found a broader Trump administration ban against Anthropic unlawful. Friday's Pentagon-specific ban, based on procurement regulations, remains in effect regardless.

Reactions

Michael celebrated the ruling on X, saying the decision "smashed Anthropic arguments" and that troops would rest easier without private companies in the chain of command. An Anthropic spokesperson said the company disagreed with the ruling, noting a separate court already found the government's parallel designation unlawful, and said it was considering further options.

The Bigger Picture

Since the ban, competitors OpenAI, xAI, Google and Microsoft have all signed Pentagon agreements. Anthropic, meanwhile, is reportedly heading toward a blockbuster IPO valued near $1 trillion, even as the White House has criticized CEO Dario Amodei's repeated warnings about AI risks.