Anthropic refuses Pentagon’s new terms, standing firm on lethal autonomous weapons and mass surveillance

In a dramatic escalation of tensions between Silicon Valley and the Pentagon, Anthropic has publicly refused the Department of Defense's ultimatum for unrestricted access to its AI systems. CEO Dario Amodei's statement, issued less than 24 hours before the Friday deadline, firmly rejects Defense Secretary Pete Hegseth's demand to renegotiate all AI contracts, drawing a clear ethical line. While expressing support for using AI to defend democracies, Anthropic explicitly opposes two applications: mass domestic surveillance and fully autonomous lethal weapons systems. This refusal sets Anthropic apart from competitors OpenAI and xAI, which reportedly acquiesced to the Pentagon's new terms, and led to Amodei being summoned for a high-stakes meeting with Hegseth at the White House.

The technical and political implications are significant. Amodei argued that 'today, frontier AI systems are simply not reliable enough to power fully autonomous weapons,' though he left the door open for future use if technology improves. In response, the Pentagon has reportedly begun assessing major defense contractors' dependence on Claude AI, a first step toward potentially designating Anthropic a 'supply chain risk'—a classification typically reserved for national security threats. The government is also considering invoking the Defense Production Act to force compliance. Amodei stated that threats 'do not change our position' and offered to facilitate a transition to another provider if the DoD chooses to offboard Anthropic, signaling the company's preparedness for a protracted standoff that could reshape the military-industrial-AI complex.