WASHINGTON: The Pentagon has imposed a formal supply-chain risk designation on American artificial intelligence company Anthropic, effectively barring its technology from use in U.S. military contracts. The decision, confirmed by Anthropic on Thursday, limits the deployment of the company’s Claude AI model in defense projects, including operations in Iran.
Read More: Anthropic refuses to bend to Pentagon on AI safeguards
Anthropic CEO Dario Amodei clarified that the designation has a “narrow scope,” applying only to Pentagon contracts. Companies can continue using Claude for other non-defense projects. “It plainly applies only to the use of Claude by customers as a direct part of contracts with the Department of War, not all use of Claude,” Amodei said, referring to the Pentagon.
View this post on Instagram
The designation follows a months-long dispute between Anthropic and the Defense Department over the company’s internal safeguards. The Pentagon reportedly viewed some of these restrictions as overly limiting for military applications. Amodei said the company plans to challenge the designation in court and has discussed ways to work with the military without dismantling its safeguards.
The announcement comes amid tensions over an internal memo by Amodei that suggested Pentagon officials were critical of Anthropic partly due to political reasons. The memo’s publication raised concerns among investors.
Anthropic has been one of the earliest U.S. AI companies to collaborate with national security agencies. Its Claude AI is reportedly used for intelligence analysis and operational planning, including for military efforts in Iran. The restriction marks an extraordinary rebuke, a status previously reserved for foreign adversaries, such as the 2019 action against Chinese tech giant Huawei.
Read More: OpenAI becomes world’s most valuable startup at $500B
Microsoft, a partner and investor in Anthropic, confirmed that Claude remains available for non-defense projects through platforms like M365, GitHub, and Microsoft AI Foundry. Amazon, another investor and user of Claude, did not respond immediately.
The Pentagon has yet to comment publicly on the designation. The move underscores growing U.S. scrutiny over the security and ethical deployment of advanced AI technologies in defense applications.