Pentagon stops using Anthropic AI tools months after blacklisting

The United States Department of Defence has completely halted its use of artificial intelligence tools developed by Anthropic. This decision comes months after the military blacklisted the prominent AI firm by labelling it a national security supply chain risk. A defence official confirmed the cessation of use in a statement released on Monday, bringing an end to a complicated period of transition and legal battles between the government and the technology company.

Back in February, Defence Secretary Pete Hegseth publicly criticised Anthropic and its chief executive, Dario Amodei, before designating the firm a supply chain risk. At the time, Hegseth announced that the Pentagon would fully phase out its reliance on Anthropic products by late August. However, multiple sources familiar with the matter revealed that the transition experienced significant delays, with the Claude model remaining operational in various military and intelligence capacities well past the initial deadline.

Insiders familiar with the situation indicated that as recently as the previous week, Claude was actively assisting with research, data analysis, intelligence gathering, and even military operations targeting Iran. The model was deeply embedded within the Pentagon's primary data organisation platform, known as the Maven Smart System, which is operated by Palantir. Because the AI technology served as a critical tool for sorting through vast quantities of information, removing it proved to be a complex technical challenge.

Experts noted that the difficulty in swiftly eliminating Claude from defence infrastructure highlights the deep integration of modern artificial intelligence within military frameworks. Lauren Kahn, a senior research analyst at Georgetown's Center for Security and Emerging Technology and a former US defence official, explained that advanced software systems are not simply plug-and-play components. Once an AI model becomes thoroughly woven into operational platforms, extracting it requires considerable time and technical effort.

The friction between the Pentagon and Anthropic began when military officials pressured the company to remove its core safety guardrails. The government demanded unfettered access to Claude for military applications, but Anthropic refused, citing serious concerns regarding mass surveillance and autonomous weapons development. In response to this refusal, the Pentagon applied the supply chain risk designation, a legal measure typically reserved for foreign companies perceived as threats to national security.

Anthropic strongly condemned the government's actions, calling the designation unprecedented and unlawful, and subsequently filed a lawsuit against the Trump administration to overturn the decision. As this legal conflict unfolds in the courts, the Department of Defence has actively formed partnerships with other major artificial intelligence companies. Competitors such as Google, xAI, and particularly OpenAI have seen their tools gain wider adoption across various military departments to fill the vacuum left by Anthropic's departure.