Pentagon and Anthropic Disagree on Use of AI in Autonomous Weapons

A top Pentagon official, Emil Michael, who is the chief technology officer, has revealed a significant disagreement with AI company Anthropic regarding the military use of its AI system, Claude. The dispute emerged over plans for integrating AI into the U.S. President’s Golden Dome missile defence programme and the potential use of AI for autonomous drones, underwater vehicles, and other defence machinery. Michael described Anthropic’s ethical restrictions on the use of its AI as an obstacle to military innovation, while Anthropic emphasised that its technology should not be used for mass surveillance of Americans or fully autonomous weapons. As a result of this standoff, the Pentagon has labelled Anthropic as a supply chain risk, which halts its defence-related contracts. Anthropic has stated its intention to contest this designation legally. Michael explained that other AI companies like Google, OpenAI, and xAI have agreed to Pentagon’s terms allowing all lawful use of AI, though Anthropic resisted, citing concerns over reliability and ethical use. The disagreement also highlights the broader debate on AI in warfare and the challenges of aligning private tech company policies with national defence requirements. The matter may now proceed to court.