U.S. Defense Secretary Summons Anthropic CEO, Pressures Military Use of AI Claude

Summoned on the 24th, expected to deliver a final ultimatum

United States Secretary of Defense Pete Hegseth has summoned the chief executive officer (CEO) of Anthropic, the developer of the artificial intelligence (AI) model "Claude," to the Pentagon. He is expected to pressure the company to enable the military use of Claude.

On the 23rd (local time), U.S. political outlet Axios, citing sources, reported that Secretary Hegseth has convened an emergency meeting on the morning of the 24th to discuss the conditions for military use of Claude and has summoned Anthropic CEO Dario Amodei to the Department of Defense.

Claude is currently considered virtually the only AI that can be used within U.S. military classified systems. The U.S. Department of Defense is demanding that Anthropic fully open up the authority to use Claude for military purposes. Rather than requiring the developer’s approval on a case-by-case basis, the company must guarantee that the AI model can be used for all lawful purposes.

The Pentagon is reported to have already used Claude last month in the operation to capture and extradite Venezuelan President Nicolas Maduro. Anthropic maintains that its technology must not be used for large-scale surveillance of U.S. citizens or for the development of fully autonomous lethal weapons that operate beyond human control.

Founded by former OpenAI employees who pursued the non-profit development of AI, Anthropic has long advocated for "safe and ethical AI." It has also maintained guidelines restricting its AI models, including Claude, from being used directly in the development of lethal weapons or in military operations.

The Department of Defense has already increased pressure by warning that if Anthropic does not agree to military use, it could designate the company as a "supply chain risk" entity. Once designated as a supply chain risk company, every contractor and supplier doing business with the Pentagon would have to certify that they do not use Anthropic’s AI model Claude. A measure typically applied to companies from adversarial nations such as China is now being aimed at a domestic firm.

In addition, the Pentagon is reportedly discussing several policy tools to force Anthropic to yield. At the meeting on the 24th, Secretary Hegseth is expected to effectively deliver a final ultimatum.

Pete Hegseth, U.S. Secretary of Defense, is answering reporters' questions at a joint press conference after concluding the 57th South Korea?U.S. Security Consultative Meeting (SCM) at the Ministry of National Defense building in Yongsan-gu, Seoul, on the afternoon of the 4th. Photo by Joint Press Corps