Pentagon draws scrutiny with Anthropic threats, Defense Production Act

The U.S. Department of Defense warned artificial intelligence company Anthropic on Tuesday that it intends to invoke the Defense Production Act (DPA) in Washington, D.C., to resolve a dispute over the company’s restrictive usage policies on its AI tools, a move aimed at compelling cooperation with federal requirements. The Defense Department issued this stern warning following a deadlock regarding the constraints Anthropic places on its large language models, which the military argues hinder necessary testing and deployment for national security applications. By threatening to utilize the DPA, a law historically reserved for prioritizing manufacturing and supply chains during crises—such as the COVID-19 pandemic—the government is signaling a shift in how it intends to secure access to critical technology. Legal experts and industry analysts have quickly characterized the Pentagon’s approach as highly unusual and potentially overreaching. They argue that while the DPA allows the government to mandate the production of goods, leveraging it to dictate terms of service or bypass safety guardrails established by a private company sets a contentious precedent. The move has drawn scrutiny because it essentially uses the weight of a national emergency law to settle a commercial and policy disagreement, raising concerns about the autonomy of private AI developers in the face of federal demands. Observers note that this tactic could escalate tensions between Silicon Valley and the federal government, particularly regarding ethical standards in AI development. This standoff underscores the growing friction between the rapid advancement of artificial intelligence capabilities and the regulatory frameworks designed to govern them. As the DoD races to integrate cutting-edge AI into its operations, companies like Anthropic have implemented strict restrictions to prevent misuse and mitigate potential existential risks associated with generative models. The threatened use of the DPA suggests that the administration is willing to use heavy-handed legal tactics to bridge this gap, potentially reshaping the relationship between the tech sector and the national security apparatus.