Pentagon and Anthropic reportedly resume talks over Claude military use

Anthropic AI

Anthropic CEO Dario Amodei is back in negotiations with the US Department of Defense after talks over military access to the company’s Claude models collapsed last week, according to a report by The Financial Times.

Amodei is now engaged in discussions with Emil Michael, the under-secretary of defense for research and engineering, in what has been described as a final effort to salvage a deal governing how the Pentagon can use Anthropic’s AI systems.

Negotiations broke down on Friday. Shortly after, Donald Trump directed federal agencies to stop using Anthropic’s tools, while Defense Secretary Pete Hegseth said the company would be designated a supply-chain risk to national security — a classification typically reserved for foreign entities considered threats to US infrastructure.

The standoff followed weeks of tension over how the military could deploy Claude. Anthropic had secured a $200 million Department of Defense contract, becoming the first major AI model provider integrated into the government’s classified networks. However, the company later sought formal guarantees that its technology would not be used for domestic mass surveillance or autonomous weapons systems. The Pentagon’s position was that the military must retain the right to use the tools for any lawful purpose.

In a memo to staff reportedly seen by the Financial Times, Amodei said the Defense Department had offered to accept Anthropic’s proposed terms if the company removed a specific phrase concerning the “analysis of bulk acquired data”. According to Amodei, that language directly addressed the very surveillance scenario the company was most concerned about. He also criticised public messaging from the Pentagon and OpenAI, describing it as misleading.

The situation intensified when OpenAI struck a new agreement with the Defense Department within hours of the White House publicly criticising Anthropic. The timing prompted backlash across parts of the tech community. Claude reportedly saw a spike in app downloads, while ChatGPT experienced a surge in uninstalls, according to analytics circulating online.

OpenAI CEO Sam Altman later acknowledged that his company “shouldn’t have rushed” its Pentagon deal and outlined revisions to its safeguards governing defence-related usage. In a post on X, Altman said Anthropic should not be designated a supply-chain risk and expressed hope that the Defense Department would offer it similar contractual terms.

Anthropic, founded in 2021 by former OpenAI researchers who split over disagreements about safety and governance, has long positioned itself as a safety-focused alternative in the AI race. That posture has drawn criticism from some government officials who argue the company has been overly restrictive in its interpretation of AI risk.

The Defense Department and Anthropic have not publicly commented on the renewed negotiations.