A Pentagon spokesperson said on Monday that the department had “stopped using Anthropic products”, BBC News reported. According to the source, this happened months after the February decision by U.S. Defense Secretary Pete Hegseth.
Hegseth then called Anthropic a supply chain risk from a national security perspective and announced that the Pentagon would stop using its products by the end of August. The BBC writes that it is unclear why the removal of the tools was delayed.
Several people familiar with the situation told the BBC that as recently as last week, Claude was being used in research, analysis, intelligence gathering, and military operations against Iran. According to them, the model was considered an important tool for processing large amounts of information.
The sources also said that Claude was integrated into Maven Smart System — a platform from the U.S. company Palantir, which is used by the Pentagon to organize intelligence and other data. Analysts and other staff used Maven to work with satellite imagery and drone data, and then uploaded them into Claude and other large language models.
Anthropic, according to the BBC, declined to comment on the Pentagon’s statement. The company previously challenged the department’s decision in court and called it “unprecedented and unlawful” after the Pentagon demanded that the protective restrictions be removed from Claude and that the military be given unrestricted access.
Amid this dispute, the department signed contracts with Google, xAI, and OpenAI, and OpenAI tools, according to two BBC sources, have become more widely used in some units in recent months. The sources also claimed that Claude was deeply integrated into Maven, so removing the system quickly would have been difficult.
Sources
