U.S. Defense Secretary Pete Hegseth has summoned Dario Amodei, the chief executive of Anthropic, to the Pentagon on Tuesday for a high-stakes meeting over the use of Claude artificial intelligence (AI) model for classified military systems.
Anthropic, which is the first company to have been authorized to work on U.S. military systems, had entered into a $200 million pilot contract with the Pentagon last year.
However, the contract was put on re-negotiations after a memo released from the office of Hegseth on January 9 this year, asked AI companies to release restrictions on their AI models for classified military use.
Claude is widely regarded inside the Pentagon as the most capable AI model available for sensitive intelligence and defense work. Losing access would be costly and disruptive. Still, Defense officials say patience has worn thin over Anthropic’s refusal to fully relax its usage safeguards.
Anthropic has offered to ease some restrictions but continues to block two areas it considers non-negotiable: mass surveillance of Americans and the development of weapons that can operate without direct human control. The company argues that these limits are essential to responsible AI use, even in national security contexts.
Pentagon leaders strongly disagree, insisting that AI systems supplied to the military must be available for all lawful missions without requiring company approval on a case-by-case basis. Officials say negotiations have stalled and are close to collapse.
According to senior Defense sources, Hegseth is expected to present Amodei with an ultimatum. The Pentagon has discussed labeling Anthropic a “supply chain risk,” a move that could cancel existing contracts and pressure other defense contractors to stop using Claude altogether.
The dispute reflects a deeper clash between Silicon Valley’s safety-first mindset and a Pentagon leadership pushing aggressively to expand AI’s role in warfare and intelligence. How that conflict is resolved could shape the future of military AI for years to come.
Also Read: Anthropic Accuses Chinese AI Companies of Illicit Distillation
