"""

Pentagon says use has ended

Anthropic AI tools are no longer being used at the U.S. Department of Defense. A Pentagon official told the BBC in writing that the department has fully discontinued the company’s products. The decision came months after the company was designated a supply-chain risk on national security grounds.

Defense Secretary Pete Hegseth announced in February that Anthropic’s use would end no later than the end of August. No explanation was given for the delay. A company spokesperson declined to comment on the development.

Why couldn’t Claude be removed right away?

According to several people familiar with the matter, Anthropic’s Claude model continued to be used until last week in research, analysis and intelligence-gathering activities. The sources said the tool was also used in military operations targeting Iran and played a critical role in sorting through large volumes of data.

Maven integration made the transition difficult

Claude was said to be embedded in the Maven Smart System, which is operated by Palantir. The Pentagon’s main platform for organizing intelligence and other data collects visual material such as satellite imagery and drone footage and feeds it into large language models.

Within this workflow, the system was used to identify potential targets and prepare short briefings sent up the chain of command. An analyst at Georgetown University said Claude was not easy to remove quickly from the platform because of how deeply it had been integrated.

New contracts shifted to rivals

As its legal dispute with Anthropic continues, the Pentagon has turned to other companies for new AI purchases. The contracts signed by the department include:

  • Google
  • xAI
  • OpenAI

According to two sources, OpenAI tools have been adopted more broadly in some military units in recent months. As a result, the distribution of public contracts in the defense technology market has shifted, while Anthropic’s room for growth on the government side has narrowed.

Anthropic was one of the first advanced AI companies to be used by government agencies handling classified work since 2024. But the company’s refusal to lift security barriers and allow unrestricted access to its products deepened the conflict with Washington; Anthropic sued the Trump administration to overturn that classification.

"""