An official says the DOD has stopped using Anthropic's tools; sources: Claude was in use as recently as last week, including in military operations against Iran

The Pentagon's reliance on deeply integrated AI tools means that removing a vendor like Anthropic can be a lengthy and complex process, impacting immediate operational capabilities.

The U.S. Department of Defense has officially stopped using artificial intelligence tools developed by Anthropic, according to a statement from a department official. This decision follows Secretary of Defense Pete Hegseth's designation of Anthropic as a national security supply chain risk in February, with an initial target to cease usage by late August. Despite this directive, sources indicate that Anthropic's Claude AI was still actively employed as recently as last week for various functions, including military operations against Iran, intelligence gathering, research, and analysis. The Pentagon's integration of Claude into systems like Palantir's Maven Smart System reportedly made its removal complex and time-consuming. This move comes as the Pentagon has increased its contracts with other AI providers, such as Google, xAI, and OpenAI, while Anthropic is currently engaged in a legal battle with the U.S. government over its designation.

The Pentagon's prolonged use of Anthropic's Claude AI, even after designating the company a security risk, highlights the significant challenge of disentangling critical AI capabilities from established defense systems. The integration into platforms like Palantir's Maven indicates how difficult it can be to quickly sever ties with technology providers once they are embedded in complex operational workflows.

This situation underscores the growing pains of the defense sector as it rapidly adopts advanced AI. It suggests that future AI procurements and integrations will need more robust off-boarding strategies and contingency planning to avoid operational disruptions and to ensure national security aligns with technological dependency.