New Delhi: Anthropic has suffered a fresh legal setback in its fight with the US government after a federal appeals court on Friday allowed the Pentagon to keep the AI company on its national security blacklist. The 2-1 ruling means the Pentagon can continue excluding Anthropic’s Claude AI from its supply chain and blocking its use by military contractors.
The case centres on Claude, Anthropic’s artificial intelligence system used for tasks including writing, coding and analysing information. Anthropic, a US AI company founded with a strong focus on AI safety, has placed restrictions on how its models can be used, including a refusal to allow Claude to be used for fully autonomous weapons or domestic mass surveillance.
Those restrictions became the centre of a dispute with the Pentagon after the two sides failed to agree on the terms for military use of Claude. The Pentagon wanted the AI to be available for any lawful military purpose, while Anthropic wanted safeguards covering uses it considered especially dangerous.
The disagreement eventually moved beyond a contract dispute when the Pentagon designated Anthropic as a national security supply chain risk in March. That action effectively cut Anthropic out of Pentagon procurement and prevented defence contractors from using its technology in their work for the department.
The appeals court found that the Pentagon had enough evidence to regard Anthropic’s restrictions as a potential national security problem. The majority said there was a legitimate concern that Claude might refuse to perform a task required during a military operation because of restrictions built into the model.
According to CNBC, Judge Gregory Katsas, a judge on the US Court of Appeals for the District of Columbia Circuit appointed by the US president, Donald Trump, in 2017, wrote the majority opinion upholding the Pentagon’s action against Anthropic. Judge Neomi Rao joined his opinion, while Judge Karen LeCraft Henderson opposed the ruling.
Anthropic has rejected the Pentagon’s reasoning and said its safeguards are meant to prevent harmful applications of AI, not interfere with military strategy. The company has argued that AI technology should not be used for autonomous lethal attacks or mass surveillance of Americans.
The legal battle is particularly notable because Anthropic and the Pentagon were previously partners. Anthropic signed a $200 million contract with the Pentagon in July 2025, and the company later worked with the government on plans to use Claude in military and classified systems before negotiations over its restrictions collapsed.
The government’s action against Anthropic involved two separate legal designations, leading to two different court battles. While Friday’s ruling upheld one of them, a federal judge in San Francisco last month struck down the other designation and found that the administration had unlawfully retaliated against Anthropic over its position on AI safety.
That means Anthropic has won one important ruling against the government but lost the separate case decided in Washington on Friday. The company said it disagrees with the latest decision and is considering further legal review.
The appeals court has not made the ruling immediately effective, giving Anthropic time to challenge it further. The company can seek a rehearing before the same panel, ask for an en banc review by the wider appeals court, or petition the US Supreme Court to take up the case.
The dispute has also become part of a wider clash between Anthropic CEO Dario Amodei and Donald Trump. Trump has repeatedly criticised Amodei, including after the CEO called for an industry-wide slowdown in AI development, adding another layer to a fight that began over how much control an AI company should have over the military use of its technology.