Pentagon ends Anthropic AI use after months of overlap with military operations
The US Department of Defence says it has stopped using Anthropic’s AI, months after Pete Hegseth designated the company a national security supply-chain risk. Sources told the BBC Claude was still being used for intelligence work and operations against Iran as recently as the previous week; its integration into Palantir’s Maven system complicated removal. The dispute centres on Anthropic’s refusal to drop safeguards against mass surveillance and autonomous weapons.
Bottom line — OpenAI, Google and xAI have Pentagon contracts, but the BBC reports OpenAI tools have gained wider use in some military departments.
Go deeper 5
-
The BBC reports that Claude had been used by the US government and military since 2024, including in classified work.
-
According to the BBC, Maven organises intelligence and visual data such as satellite imagery and drone footage; sources said analysts used Claude to process information and prepare briefings.
-
Georgetown researcher Lauren Kahn told the BBC that removing Claude was difficult because it was deeply integrated into Maven: AI tools are not simply plug-and-play.
-
Anthropic called the designation “unprecedented and unlawful” and sued the Trump administration, the BBC reports.
-
The BBC says Hegseth set a six-month phase-out deadline on 27 February; the Pentagon has not explained why use continued beyond the planned deadline.