Pentagon vs. Anthropic: Michael on AI Standoff

The Pentagon's Under Secretary Emil Michael strongly criticized AI developer Anthropic for halting negotiations over military use of its Claude AI, calling their CEO a 'liar with a god complex' and accusing the company of a PR stunt.

Feb 27 at 5:27 PM2 min read
Image representing a standoff between the Pentagon and an AI company, possibly with a visual metaphor for AI models or data flow.

A high-stakes disagreement is unfolding between the Pentagon and AI developer Anthropic, particularly concerning the military's potential use of its Claude large language model. Under Secretary of War Emil Michael recently escalated the rhetoric, criticizing Anthropic's CEO for what he called a 'god complex' and a 'liar' during an interview with the Bloomberg Podcast, after the AI firm reportedly broke off negotiations.

The core of the dispute revolves around Anthropic's demands for explicit contractual language prohibiting its AI, including Claude, from being used for surveillance of U.S. citizens or autonomous lethal strikes without human oversight. Michael contends that the Pentagon had already conceded to these points in writing, affirming adherence to all applicable laws and regulations such as the National Security Act of 1947 and the FISA Act, which already forbid mass surveillance of Americans.

Pentagon's Stance on AI Safeguards

Michael emphasized that the Department of Defense's proposal included commitments to human oversight across all stages of autonomous weapons development, engagement, and use. He expressed surprise at Anthropic's decision to halt talks, especially given what he described as minor remaining differences and extensive good-faith negotiations over three months. This unexpected move, he suggested, appeared to be part of a pre-planned public relations strategy by Anthropic.

Pentagon’s Emil Michael on Anthropic Talks, Military’s AI Use | Bloomberg Talks — from Bloomberg Podcast

The Under Secretary also highlighted the Pentagon's broader strategy of fostering diverse AI partnerships. He noted that the Department of Defense is actively engaging with multiple technology companies, including Elon Musk's XAI for classified and unclassified networks, to ensure a range of options for military applications. This approach aims to prevent dependency on a single vendor and leverage the best available AI tools to protect warfighters and enhance operational efficiency and lethality.

Michael stressed the critical role of AI, including advanced models like Claude, in modern national security. He pointed to evolving threats like drone swarms and hypersonic missiles, where rapid response times are crucial. The Pentagon's priority, he stated, is to equip its forces with the most effective tools for defense, while always maintaining human oversight and strict adherence to legal frameworks, adapting to the changing landscape of warfare.