Anthropic faces Pentagon ultimatum over AI model access

On Tuesday morning, US Defense Secretary Pete Hegseth met with Anthropic CEO Dario Amodei to deliver an ultimatum. According to Axios, Anthropic has until Friday evening to provide unrestricted access to its AI model or face consequences. Hegseth communicated that the Pentagon will either declare Anthropic a “supply chain risk”—a designation typically reserved for foreign adversaries—or invoke the Defense Production Act to force the company to tailor its model to military needs.

The dispute highlights growing ideological friction within the administration. David Sacks, the administration’s AI czar, has publicly criticized Anthropic’s safety policies as “woke.” Using the Defense Production Act—a US law allowing the president to force production for national defense—to resolve a dispute over AI guardrails would mark a significant expansion of the law’s modern application. Typically, the law is reserved for national defense manufacturing, such as compelling companies to produce ventilators and masks during emergencies. According to Dean Ball, a senior fellow at the Foundation for American Innovation and former senior policy advisor on AI in Trump’s White House, this aggressive use of the law would essentially be the government threatening to put a company out of business over political disagreements, which could signal to investors that the US is no longer a stable place to do business.

This standoff represents a high-stakes conflict, as Reuters reported that Anthropic doesn’t plan on easing its usage restrictions. Anthropic has long restricted its technology from being used for mass surveillance of Americans or for fully autonomous weapons, and is refusing to compromise on these points. However, Pentagon officials argue that military technology use should be governed by US law and constitutional limits, rather than the usage policies of private contractors.

Analysts warn that the Pentagon’s aggressive posture stems from a critical lack of redundancy. Anthropic is currently the only frontier AI lab with classified Department of Defense (DOD) access. In this context, frontier AI refers to advanced AI models at the cutting edge of capability. While the Pentagon has reportedly reached a deal to use xAI’s Grok in classified systems, experts argue the military is failing to meet its own directives. Ball noted that the agency appears to be falling short of a National Security Memorandum that directs federal agencies to avoid dependence on a single classified-ready frontier AI system. “The DOD has no backups. This is a single-vendor situation here,” Ball said. He added that if Anthropic canceled the contract tomorrow, it would be a serious problem for the DOD, noting that they cannot fix the situation overnight.

Why it matters

The Pentagon’s threat to use the Defense Production Act against a private AI lab marks a significant expansion of the law’s use. It also highlights the military’s dangerous lack of redundancy in classified AI systems.