Anthropic chief executive Dario Amodei is scrambling to rescue a US defence department deal before the Pentagon slams the supply chain door in his face.
He has been in talks with the US Department of Defence’s Undersecretary of Defence for Research and Engineering, Emil Michael, to patch up a contract covering the Pentagon’s access to Anthropic’s AI models, according to people familiar with the matter.
A fresh agreement would let the US military keep using Anthropic’s technology and cut the risk of the firm being branded a supply chain risk, a threat issued by defence secretary Pete Hegseth on Friday but not yet carried out.
This late push follows the messy collapse of negotiations last week, after Michael attacked Amodei as a “liar” with a “God complex.”
Deliberations fell apart a day later when they could not agree on wording Anthropic said was needed to block AI use for mass domestic surveillance, one of its red lines alongside lethal autonomous weapons.
Amodei wrote: “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.”
Amodei wrote that messaging from the Pentagon and OpenAI was “just straight up lies about these issues or tries to confuse them”.
He suggested Anthropic had been pushed aside because “we haven’t given dictator-style praise to Trump” compared with OpenAI chief Sam Altman.
Anthropic landed a $200mn agreement with the US defence department in July 2025 and was the first AI model used in classified settings and by national security agencies.
The row intensified after the Pentagon demanded AI companies accept their tools being used for any “lawful” purpose, and Hegseth moved to threaten the supply chain label that would force military suppliers to cut ties.
