Pentagon ceases use of Anthropic AI tools, BBC reports
A department official told the BBC that the military stopped deploying the models.
The US Department of Defence has stopped using Anthropic's artificial intelligence tools, according to a statement given by a department official to the BBC. The decision follows months after Defence Secretary Pete Hegseth designated the company a national security supply chain risk.
Anthropic had provided tools to the US government and military since 2024. The relationship fractured after the Pentagon requested the removal of safety guardrails on the Claude model to grant the military unrestricted access. Anthropic refused, citing concerns over mass surveillance and autonomous weaponry. In response, Pete Hegseth labelled the startup a supply chain risk and set a six-month deadline on 27 February 2026 to end its use by late August 2026. Anthropic sued the Trump administration to challenge the decision.
Despite that deadline, personnel continued using Claude models, including Mythos, for research, intelligence gathering and target identification, sources told the BBC. The software was integrated into Palantir's Maven Smart System, which processes satellite imagery and drone footage for US military analysis and international intelligence sharing. The Pentagon has since signed contracts with alternative providers, including Google, xAI and OpenAI.
An Anthropic spokesman declined to comment to the BBC regarding the Pentagon's statement. The move occurs as Anthropic CEO Dario Amodei met twice with President Donald Trump at the White House, stating that discussions regarding artificial intelligence risks remain ongoing.
Newsletter
Markets in your inbox, weekly
LATAM-focused analysis, investing ideas, and the week in finance.
Keep reading
