Anthropic’s CEO, Dario Amodei, has re-entered negotiations with the Department of Defense, aiming to mend the company’s strained ties with the US military. This move comes after the startup faced potential exclusion from defense-related projects due to being labeled a “supply chain risk.” The discussions took a nosedive last Friday following heated public disputes over Anthropic’s refusal to offer the Pentagon unrestricted access to its AI technology, a gap that competitors like OpenAI are eager to fill.

Amodei is reportedly in dialogue with Emil Michael, the under-secretary of defense for research and engineering, to establish a new agreement that would keep Anthropic’s Claude AI models in military use. This information, sourced from the Financial Times, highlights the contentious relationship between the two, exacerbated by Michael’s recent social media outbursts. He accused Amodei of being a “liar” with a “God-complex,” and claimed the CEO’s actions endangered national security.

The stakes for securing this deal are high for Anthropic. Defense Secretary Pete Hegseth considered classifying the company as a supply chain risk, a designation usually reserved for entities with foreign ties posing security threats. Such a label could reverberate through the US tech industry, compelling firms to abandon Claude and cut links with Anthropic to maintain their defense contracts.

A leaked memo penned by Amodei to Anthropic employees, uncovered by The Information and verified by the FT, may further strain the company’s relationship with the Trump administration. In this memo, Amodei denounced OpenAI’s agreement with the Pentagon as mere “safety theater” and accused both parties of spreading “straight up lies.”

Amodei speculated that Anthropic’s deteriorating rapport with the government stemmed from its lack of financial contributions to Trump and its refusal to offer excessive praise. In contrast, OpenAI’s CEO, Sam Altman, along with co-founder and president Greg Brockman, have reportedly aligned themselves with Trump, with Brockman notably being a major donor.

In the memo, Amodei also mentioned that the Department of Defense was nearing an acceptance of Anthropic’s proposed terms:

“Near the end of the negotiation the [department] offered to accept our current terms if we deleted a specific phrase about ‘analysis of bulk acquired data’ which was the single line in the contract that exactly matched this scenario we were most worried about. We found that very suspicious.”

Anthropic’s ugly dispute with the Pentagon has centered around the Defense Department’s insistence on carte blanche access to the company’s technology and the startup’s refusal to compromise on its 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 oversight. Hegseth has insisted the AI technology used by the department should be available for “any lawful use,” terms Anthropic has refused amid concerns it could cross these red lines. xAI and OpenAI have reportedly agreed to those terms.