Anthropic CEO Dario Amoudi has reportedly returned to the negotiating table with the Department of Defense in an effort to save the company’s relationship with the U.S. military and prevent it from being pulled out of defense work due to “supply chain risk.” Talks between the two parties broke down on Friday after weeks of bitter public spats over the Pentagon’s refusal to grant unrestricted access to its AI, with rivals such as OpenAI rushing to fill the void.
Amoudi is in talks with Under Secretary of Defense for Research and Engineering Emil Michael about a new agreement that would allow the US military to continue using Anthropic’s cloud AI models. Financial TimesCiting unknown sources with information on the matter. Michael attacked Amodei on social media last week amid a tense standoff over the acceptable military use of AI, calling the executive a “liar” with a “God complex” and accusing him of “jeopardizing our nation’s security.”
Getting a new deal can be a matter of survival for an American startup. Defense Secretary Pat Hegseth said Friday he would designate Anthropic a supply chain risk, a category usually reserved for companies with ties to foreign governments that pose national security threats to the United States. The designation will affect the entire US tech ecosystem, forcing firms to leave the cloud and cut ties with the company if they want to continue working on defense contracts.
A newly leaked memo sent from Amody to Anthropic staff on Friday was first reported by Information Also seen by and FTis likely to inflame already strained relations between the company and the Trump administration. In it, Amoudi reportedly criticized OpenAI’s deal with the Pentagon as “safety theater” and called both sides’ messages “outright lies.”
Amudi suggested that Anthropic’s relationship with the federal government has soured because, unlike OpenAI or its executives, “we have not donated to Trump” and “we have not appreciated Trump’s authoritarian style.” OpenAI CEO Sam Altman is one of many Silicon Valley heavyweights who have been working with the president on deals since he returned to office, and cofounder Greg Brockman, also OpenAI’s longtime president, is a Trump megadonor.
In the memo, Amodi also said the Defense Department was close to accepting Anthropic’s terms:
“Near the end of the negotiations (the department) offered to accept our current terms if we deleted a specific clause about ‘analysis of bulk acquired data’ which was the only line in the contract that closely matched the scenario we were most worried about. We found that very suspicious.”
Anthropic’s ugly dispute with the Pentagon centers around the Department of Defense’s insistence on access to the company’s technology and its refusal to compromise on two red lines for military use: no mass surveillance of Americans and no lethal autonomous weapons, AI systems with the power to kill without human supervision. Hegseth has insisted that AI technology used by the department must be available for “any lawful use,” terms Anthropic has rejected amid concerns it could cross red lines. xAI and OpenAI have reportedly agreed to these terms.
