Anthropic CEO Dario Amodei has released a definitive statement regarding the company’s escalating friction with the U.S. Department of War. While Anthropic has successfully deployed its models within classified government networks for intelligence analysis, it has hit a stalemate over two specific red lines: mass domestic surveillance and the development of fully autonomous weapons. Amodei argues that today’s frontier AI systems simply lack the necessary reliability to select and engage targets without human intervention, posing an unacceptable risk to both warfighters and civilians.

The dispute has reached a boiling point, with federal officials threatening to designate the American AI firm as a “supply chain risk”—a label typically reserved for hostile foreign powers. Furthermore, the government has suggested invoking the Defense Production Act to legally compel the company to remove its built-in safety guardrails. This creates a stark contradiction where the state acknowledges the technology’s essential role in national security while simultaneously labeling its developers as a threat for prioritizing safety protocols over unrestricted military expansion.

Despite the pressure, Anthropic maintains that its refusal is rooted in a commitment to democratic values. Amodei noted that the company will not facilitate tools for mass surveillance that could bypass current legal protections for citizens. The firm expressed a preference to continue serving the military under restricted terms but remains prepared to offboard its technology if the Department refuses to respect these fundamental safety boundaries.

Anthropic CEO Dario Amodei has released a definitive statement regarding the company’s escalating friction with the U.S. Department of War. While Anthropic has successfully deployed its models within classified government networks for intelligence analysis, it has hit a stalemate over two specific red lines: mass domestic surveillance and the development of fully autonomous weapons. Amodei argues that today’s frontier AI systems simply lack the necessary reliability to select and engage targets without human intervention, posing an unacceptable risk to both warfighters and civilians.

The dispute has reached a boiling point, with federal officials threatening to designate the American AI firm as a “supply chain risk”—a label typically reserved for hostile foreign powers. Furthermore, the government has suggested invoking the Defense Production Act to legally compel the company to remove its built-in safety guardrails. This creates a stark contradiction where the state acknowledges the technology’s essential role in national security while simultaneously labeling its developers as a threat for prioritizing safety protocols over unrestricted military expansion.

Despite the pressure, Anthropic maintains that its refusal is rooted in a commitment to democratic values. Amodei noted that the company will not facilitate tools for mass surveillance that could bypass current legal protections for citizens. The firm expressed a preference to continue serving the military under restricted terms but remains prepared to offboard its technology if the Department refuses to respect these fundamental safety boundaries.