Pentagon Bans Anthropic's Claude Over Ethical Red Lines, Shifts to OpenAI and Google

The US Department of Defense has officially ceased using all Anthropic products, including the Claude AI model, according to a BBC report dated October 5.

In February, Defense Secretary Pete Hegseth designated Anthropic as a national security "supply chain risk" and ordered a phase-out within six months.

The core disagreement centered on military use restrictions. The Pentagon sought AI tools available for "any lawful purpose," but Anthropic maintained two red lines: its technology could not be used for mass surveillance or for fully autonomous lethal weapons.

In response, the Pentagon classified Anthropic as a supply chain risk — marking the first time a major AI vendor has refused its largest government customer, costing Anthropic the contract.

BBC sources indicate that as of late September, Claude was still employed for research, analysis, intelligence gathering, and military operations targeting Iran.

Claude had been integrated into the Maven Smart System, a critical intelligence platform operated by Palantir that processes satellite imagery, drone footage, and other large-scale intelligence data. These data feeds were further processed by Claude for tasks such as identifying potential military targets.

A BBC source witnessed a Palantir demonstration of Maven, noting that military officers routinely use the system to compile one-page briefings that move up the chain of command.

The deep integration with Maven complicates replacement efforts.

Meanwhile, the Pentagon has signed contracts with OpenAI, Google, and xAI. Recent months have seen broader adoption of OpenAI tools across military departments.

The transition from the February phase-out order to full discontinuation in October took over seven months.