Pentagon considers severing ties with Anthropic over AI policy

The Pentagon is considering severing ties with Anthropic over security restrictions imposed by the startup on its artificial intelligence (AI) solutions. These safeguards, which prevent the use of the Claude chatbot for mass surveillance and autonomous weapons development, have caused dissatisfaction within the U.S. Department of Defense, as revealed by Axios on October 16.

Claude is the only AI model authorized to operate in the U.S. military's classified systems. However, Anthropic's policies, which prohibit the use of AI for certain purposes, conflict with military interests. The company is willing to negotiate some terms but refuses to remove all restrictions, leading to a crisis with the Pentagon.

In light of this impasse, U.S. defense officials are exploring alternatives such as Google, OpenAI, and xAI that could provide AI for all lawful purposes. Despite this, a senior government official warned that competing AIs have not yet matched Claude's specific capabilities.

U.S. Secretary of Defense Pete Hegseth warned that Anthropic could be classified as a "supply chain risk" if the agreement is severed, which would impact its business with third parties connected to the military. Anthropic, for its part, is maintaining discussions with the Department of Defense, insisting on its commitment to the safe use of technology.

The situation between the Pentagon and Anthropic highlights the conflict between security and flexibility in using AI for military purposes. The search for a balance that satisfies both parties remains a significant challenge.