The Pentagon wants Anthropic to remove AI safety guardrails. Anthropic is refusing.

Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei a deadline: back down on AI safeguards or face removal from Department of Defense systems. The deadline was Friday.

Anthropic response: no.

According to Amodei, Anthropic cannot accede to the Pentagon requests in an AI safeguards dispute. The company is standing by its safety commitments despite threats to remove the company from DoD systems.

What the Pentagon Wants

The Pentagon demanded that Anthropic turn off safety guardrails and allow any lawful use of Claude. This would enable the military to use Claude for mass domestic surveillance and autonomous weapons systems.

According to The Guardian, the Pentagon wants unfettered access to Anthropic AI model, without the safety checks the company has built in.

What Anthropic Is Protecting

Anthropic has pushed back against allowing Claude to be used for mass domestic surveillance or in autonomous weapons systems. The company position is that these uses would violate its ethical commitments.

According to Amodei, the company strong preference is to continue serving the Department and warfighters with its requested safeguards in place. Should the Department choose to offboard Anthropic, the company will work to enable a smooth transition to another provider, avoiding any disruption to ongoing military planning and operations.

A First for Big AI

This represents one of the most public fractures between a major AI company and the U.S. military. Unlike competitors like Microsoft and Amazon who have aggressively pursued defense contracts, Anthropic has drawn a line.

The company is betting that its commitment to safety ethics will resonate with enterprise customers and the broader public. The question is whether that bet pays off or whether the Pentagon simply moves to another provider.

Sources

• Reuters: Anthropic rejects Pentagon requests in AI safeguards dispute
• CNBC: Anthropic CEO Amodei says Pentagon threats do not change our position
• The Guardian: Anthropic says it cannot in good conscience allow Pentagon to remove AI checks
• Axios: Hegseth gives Anthropic until Friday to back down on AI safeguards