Anthropic Secures U.S. Court Victory and Is Removed from Pentagon List

On Thursday night (27th), a federal judge ruled that the Pentagon's decision to classify AI developer Anthropic as a "supply chain risk" violated the law and ordered the classification removed.

Judge Rita Lin wrote that, while the military should have broad latitude to decide which companies to work with, the actions against Anthropic "constituted unlawful retaliation in violation of the First Amendment" and that the company "was denied the pre-deprivation process required by the Fifth Amendment."

The dispute between the Department of Defense and Anthropic began earlier this year, after the developer refused to remove internal safety safeguards that would prevent the military from using the Claude AI model in autonomous weapons and mass surveillance. Anthropic argued that its models were not reliable enough to be used for those purposes.

Defense Secretary Pete Hegseth stated that the government could not allow the U.S. Armed Forces to be limited or have their tools controlled by a private company.

In February, he made the unprecedented decision to classify Anthropic as a supply chain risk. The classification means that no part of the Pentagon, including contractors, could work with the developer's products. That label had previously been used only for companies with alleged ties to foreign adversaries.

Anthropic filed a lawsuit in March.

Judge rejected Anthropic's inclusion on the Pentagon list

Lin, appointed by former President Joe Biden to the Northern District of California, rejected the Pentagon's claims that the department's inability to "trust" Anthropic justified assigning the label.

"The empty invocation of national security is not a blank check to punish and retaliate against critics of the government," she wrote.

She further added that the evidence shows the Pentagon wanted to make a "public example of Anthropic for its 'arrogance' in criticizing the government, without any articulated basis to believe that Anthropic would actually sabotage its model."

In a statement, a spokesperson for the developer said they "welcome" the decision.

"We remain focused on working productively with the government to harness AI for our national security, so that all Americans benefit from this technology," the spokesperson added.

The Pentagon did not respond to a request for comment.

Judge Lin noted that other parts of the government continued to be in contact with Anthropic even after the Pentagon's action.

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

In an interview with Axios in June, President Donald Trump stated that, although he had previously considered Anthropic a threat to national security, he no longer believed that to be the case.

A second lawsuit filed by the developer related to the supply chain risk label is still under review by a court in Washington, DC.