Particle.news
Download on the App Store

Pentagon Weighs Split With Anthropic, Threatens Sanctions Over Military AI Limits

Months of stalled talks over Anthropic’s bans on violent, weapons-related, mass-surveillance uses set up this escalation.

Overview

  • Axios reporting, echoed by Heise, says the Pentagon may label Anthropic a supply chain risk, a move that could bar defense partners from using Claude.
  • The Defense Department has pressed major AI vendors to allow use for all lawful purposes, including weapons development and battlefield operations, which Anthropic has not accepted.
  • Anthropic reiterates that its policies prohibit uses involving violence, weapons work or mass surveillance and declines to confirm any role in specific operations.
  • The Wall Street Journal reported Claude was used via Palantir in the U.S. operation to capture Nicolás Maduro, though key details remain unconfirmed.
  • Axios also reports Claude is currently the only model permitted on the Pentagon’s classified systems and that officials are considering canceling a promised contract of up to $200 million, even as the department pursues work with xAI.