Secretary of Defense Declares Anthropic a Supply Chain Risk After Contract Dispute

Summary

In a dramatic escalation of tensions between the AI industry and the U.S. government, Secretary of Defense Pete Hegseth unilaterally declared Anthropic a supply chain risk via Twitter, announcing that any Defense Department contractor would be prohibited from using Anthropic's products. The declaration came just one hour after President Trump had announced what appeared to be a measured resolution—canceling Anthropic's government contract with a six-month wind-down period following the company's refusal to accept certain contract terms.

The dispute centers on Anthropic's adherence to its responsible scaling policy and refusal to provide what the company characterized as unrestricted access for mass domestic surveillance and lethal autonomous weapons. According to analyst Zvi Mowshowitz, who documented the events in detail, Anthropic declined contract terms that OpenAI had accepted, particularly around restrictions on use cases. The company maintained red lines around meaningful restrictions on domestic surveillance and autonomous weapons systems, positions that reportedly conflicted with the Defense Department's demands for broader access.

Hegseth's Twitter declaration, which bypassed normal regulatory procedures, stated that Defense contractors could not use Anthropic products in any capacity. The announcement triggered immediate market turbulence, with approximately $150 billion wiped from public markets in after-hours trading. Legal experts and industry observers questioned both the legality and implementation mechanism of such a sweeping declaration issued via social media. The move represents an unprecedented use of supply chain security authorities to pressure an AI company over contract negotiation disagreements.

The incident highlights deepening friction between AI safety-focused companies and government agencies seeking broad access to frontier AI systems. While sources indicated Trump's initial announcement was intended as a face-saving de-escalation that would allow time for further negotiation, Hegseth's subsequent action transformed the situation into what observers called an attempt at 'corporate murder with a global blast radius.' The contrast between Anthropic's approach and OpenAI's acceptance of similar government contract terms has exposed fundamental disagreements within the AI industry about balancing national security cooperation with ethical commitments and safety policies.

• The action came just one hour after President Trump announced a more measured six-month wind-down period, which was reportedly intended as a de-escalation
• The incident exposes fundamental tensions between AI safety commitments and government demands for broad access to frontier AI systems