In a tense standoff that has captured attention, Anthropic has refused the Pentagon’s demands for unrestricted access to its artificial intelligence technologies, just a day before the ultimatum deadline set by the Department of Defense.

This clash is the result of a series of public declarations, social media exchanges, and intense negotiations. The crux of the issue revolves around Defense Secretary Pete Hegseth’s push to renegotiate existing contracts with AI laboratories, seeking broader military access. While other companies like OpenAI and xAI have reportedly agreed to the new terms, Anthropic remains firm in its stance against mass surveillance of American citizens and the deployment of fully autonomous weapons without human oversight. This resistance has led to Anthropic CEO Dario Amodei being called to the White House for direct talks with Secretary Hegseth, who allegedly issued a firm ultimatum: comply by the end of the week or face consequences.

In a statement released Thursday evening, Amodei expressed his commitment to leveraging AI for national defense purposes, stating, “I deeply believe in the existential importance of using AI to protect the United States and other democracies, as well as to counter our autocratic adversaries.” He emphasized Anthropic’s proactive efforts in collaborating with the Department of Defense and intelligence agencies.

Amodei clarified that while Anthropic has never opposed specific military operations or limited the usage of its technology arbitrarily, there are specific scenarios where AI could potentially compromise democratic values. He highlighted concerns about mass domestic surveillance and the use of fully autonomous weapons, though he acknowledged that partially autonomous weapons are essential for defending democracy. He did not entirely dismiss the possibility of future alignment with the military’s use of fully autonomous weapons, but stressed that current AI systems are not yet reliable enough for such applications.

Reports indicate that the Pentagon has asked major defense contractors to evaluate their reliance on Anthropic’s AI model, Claude, a move that could pave the way for classifying the company as a “supply chain risk” — a designation typically reserved for national security threats. The Defense Department is also considering invoking the Defense Production Act to compel Anthropic’s compliance.

In his statement, Amodei remained resolute, saying, “The Pentagon’s threats do not alter our stance: we cannot, in good conscience, meet their request.” He assured that if the Department decides to disengage from Anthropic, the company will facilitate a seamless transition to another provider, ensuring no disruption to ongoing military operations and planning. Anthropic’s models, he noted, will remain accessible under the terms they have proposed for as long as necessary.