Pentagon confirms it has stopped using Anthropic’s Claude as Washington presses the company

The Pentagon is no longer using Anthropic’s AI tools, the department told the BBC¹. It “has ceased the use of Anthropic products”, months after declaring the company a supply chain risk³. Defence Secretary Pete Hegseth applied that label in February and said the Pentagon would stop using the tools, after the firm refused to remove safety guardrails from them^1,2.

The designation has survived legal challenge. A federal appeals court upheld the Department of Defense’s exclusion of Anthropic’s Claude models in the case brought by Anthropic PBC against the US Department of War⁶, and the company lost its bid to overturn the national security designation⁷. Judge Gregory Katsas wrote the holding, in a ruling critics argue gets free speech wrong⁸. The dispute turned on the Pentagon’s demand that Anthropic drop its restrictions on military use⁸. The company has continued to expand into other federal agencies even as the Pentagon stays off Claude⁵.

Why it matters to investors

Anthropic is expected to go public despite market uncertainty and calls for an AI slowdown, with its plans muddled by questions about regulation and safety risk¹⁰. The company has told investors that the US government is a problem and that it may experience business disruptions¹¹. Senator Bernie Moreno’s letter landed ahead of what Axios calls an expected historic IPO, with valuation estimates as high as nearly $2 trillion⁹.

The political pressure is explicit. Moreno, a Republican from Ohio, urged chief executive Dario Amodei against an “alarmist approach” and “panic-inducing commentary” on AI in a letter sent on Monday⁹. Axios describes it as the latest example of President Trump’s allies targeting Amodei⁹, whose call for the sector to curb its development pace followed rising public anxiety about frontier safety¹¹.

What to watch

The first question is scope. The Pentagon’s exclusion is now court-backed^6,7, so watch whether other departments follow or whether the civil-agency expansion continues⁵. Government contractors are reading the ruling for what a supply-chain-risk designation means for their own usage restrictions⁶.

The second is disclosure. If the company is already warning investors about government risk and possible business disruption¹¹, that language will appear in any prospectus alongside the regulation and safety questions already flagged¹⁰. A separate class action filed by a Claude user over biometric scans adds another litigation line to track⁷.