Judge Stops Pentagon From Blacklisting AI Maker Anthropic

A federal judge in California has blocked the Pentagon from blacklisting Anthropic, the company behind the Claude AI models. Judge Rita Lin ruled that Defense Secretary Pete Hegseth's decision to label Anthropic a "supply-chain risk" was unconstitutional retaliation. That label would have banned the company from federal contracts. The judge also removed similar sanctions from nine government agencies, calling the punishment "arbitrary and capricious." The Pentagon is expected to appeal.

The fight began after a $200 million military deal. Anthropic had agreed to let the Pentagon use Claude, but the company insisted on limits. It refused to support lethal autonomous weapons or mass surveillance systems. The Pentagon rejected those limits, saying a contractor cannot control how its technology is used after it is delivered. When talks broke down in February, the Pentagon blacklisted Anthropic, which would also prevent military contractors from doing business with the company.

Judge Lin saw it differently. She noted that the government was still discussing using Anthropic's new model in sensitive projects. That, she wrote, is not consistent with a genuine fear that Anthropic is a saboteur who would poison its software. The ruling says the Pentagon may still choose a different AI vendor, but it cannot punish Anthropic for setting ethical boundaries. Anthropic is one of the world's most advanced AI labs, and a separate lawsuit in Washington, DC, is still moving forward.

If the ruling stands, it could give AI companies more confidence to negotiate limits with the government. It also sends a message: the government cannot use blacklists to pressure companies into dropping safety standards. For everyday people, that matters because it shapes how AI will be used by the military—and whether companies can say no to dangerous uses without being crushed.