Anthropic Loses Pentagon Fight As US Appeals Court Upholds Military Supply Chain Ban

A US appeals court on Friday upheld the Pentagon’s decision to exclude AI company Anthropic from its military supply chain after the company refused to remove safeguards on its Claude models. The ruling keeps in place a supply chain risk designation that cancelled Anthropic’s military contracts and barred Pentagon contractors from using its technology.

The dispute began in February when Anthropic rejected Pentagon Secretary Pete Hegseth’s demand that it remove restrictions on Claude’s use for fully autonomous lethal weapons and mass domestic surveillance of Americans. The Pentagon formally designated Anthropic a supply chain risk in March under federal procurement rules.

The military had been using Claude across classified and other sensitive systems. According to the appeals court opinion, an Anthropic executive had questioned the use of Claude by Pentagon contractor Palantir during the US military operation that captured Venezuelan President Nicolas Maduro on January 3.

That objection “led to alarm” within the Pentagon and raised concerns about whether Anthropic could cause its software to stop working or trigger another disastrous action that could endanger US warfighters, Under Secretary of Defense Emil Michael said, according to the court opinion.

Anthropic has argued that it cannot modify its models once they have been delivered to the military. However, the Washington appeals court found that the company controls the behaviour of each new version of Claude and that the military must keep up with the latest offerings.

Judge Gregory Katsas, writing for two of the three judges, said that “in our Republic, it is the President and the Secretary of War who must determine how best to balance” the relevant risks. The court upheld the Pentagon’s action under public procurement regulations.

The designation was an unusual use of government authority against a major US technology company. The Pentagon quickly turned to competitors. OpenAI, the creator of ChatGPT, signed an agreement with the Pentagon on the same day as Anthropic’s ban, while Elon Musk’s xAI signed in February. Google and Microsoft followed with agreements in the subsequent months.

The dispute has also put Anthropic at odds with the White House despite the company’s major role in the AI industry. Anthropic, valued at nearly $1 trillion, is reportedly heading towards a major IPO expected within weeks. CEO Dario Amodei has repeatedly warned about the dangers posed by AI, while President Donald Trump has dismissed such warnings as a “hoax”.

Under Secretary Michael celebrated Friday’s ruling on X, writing: “The hammer of justice has smashed Anthropic arguments.” He added that warfighters would sleep better knowing that no private company could “insert their opinions in the chain of command.”

The ruling conflicts with a separate federal court decision in California last month, which found a broader Trump administration ban against Anthropic unlawful. The Pentagon’s action upheld on Friday was based on different public procurement regulations and therefore remains in effect.

Anthropic said it “respectfully disagrees” with the decision. A company spokesperson said another federal court had already found the government’s parallel designation unlawful and that Anthropic remained confident in its position while considering all options, including further review.