AI firm Anthropic wins case challenging Pentagon blacklisting

The artificial intelligence company Anthropic won a federal lawsuit challenging the Trump administration's decision to brand the company a national security risk earlier this year, blocking it from doing business with the entire government and military contractors.

A federal judge in San Francisco ruled Thursday that the government's actions - justified largely by a four-page memo - were "illegal and baseless."

Anthropic's leaders clashed with top defense officials in February over the Pentagon's demand that it must be allowed to use the company's AI models for any lawful purpose. The company sought guarantees that its systems would not be used to power fully autonomous weapons or carry out mass domestic surveillance.

Defense officials refused, and labeled the company a supply chain risk - a designation with far reaching consequences for Anthropic's business that the judge ruled threatened to "cripple" the company by shutting it out of federal contracts.

U.S. District Judge Rita F. Lin in San Francisco ruled Thursday that while the Pentagon was free to pick and choose which contractors it works with, the administration had no basis for taking such sweeping actions.

Officials' "contemporaneous words and deeds confirm that the challenged actions were based on a desire to make a public example out of Anthropic for its 'arrogance' in criticizing the government," Lin wrote.

Anthropic makes the ChatGPT rival Claude, which has over the past year become popular with software companies for its coding skills. Thursday's ruling follows a preliminary order by the judge in March and represents a stinging rebuke of the Trump administration's handling of the situation. But a separate but overlapping case in Washington is still playing out.

The government's lawyers said at a recent hearing in the case that the Pentagon was continuing to remove Anthropic's tools from its systems, and expected to complete the work by the end of September.

Anthropic said on Thursday that it welcomed the judge's decision.

"We remain focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology," the company said in a statement.

The Pentagon and the White House did not immediately respond to a request for comment.

Anthropic was the first AI company to be cleared to work on classified government systems in 2024 and initially worked enthusiastically with defense officials. When the battle between the two sides reached its climax in late February, it appeared that the relationship between one of the nation's leading AI companies and the federal government would be left badly damaged.

The Trump administration's stance softened in April, when Anthropic announced it had developed a new model called Mythos that had powerful new computer hacking abilities. Federal officials raced to get their hands on the technology to understand the risks it posed, and Anthropic chief executive Dario Amodei visited the White House.

Lin wrote in her decision Thursday that the government's ongoing discussions with Anthropic about using Mythos undercut its argument that the company posed a risk.

"None of that is consistent with a genuine fear that Anthropic is a saboteur who would poison its software to harm national security," she wrote.