United States Secretary of Defense Pete Hegseth issued a formal directive on Friday designating the artificial intelligence startup Anthropic as a supply-chain risk, a move that effectively prohibits any entity doing business with the U.S. military from engaging in commercial activity with the company. The announcement, delivered via social media and official channels, has triggered significant disruption within the defense technology sector and Silicon Valley, as thousands of federal contractors and subcontractors evaluate their reliance on Anthropic’s Claude AI models. The designation marks a dramatic escalation in a long-standing dispute between the Department of Defense and one of the nation’s leading AI developers regarding the ethical boundaries of military technology.
Secretary Hegseth’s directive was explicit in its scope, stating that the prohibition is effective immediately and applies to all contractors, suppliers, and partners of the United States military. The decision follows weeks of increasingly friction-filled negotiations concerning the terms under which the Pentagon could utilize Anthropic’s Large Language Models (LLMs). While Anthropic has sought to establish specific guardrails against the use of its technology for mass domestic surveillance and fully autonomous lethal weaponry, the Department of Defense has maintained that any agreement must permit the military to apply AI to all lawful uses without external restrictions.
The Statutory Basis for the Designation
The Department of Defense utilized its authority under 10 U.S.C. § 3252 to label Anthropic a supply-chain risk. This specific statute allows the Secretary of Defense to exclude or restrict certain vendors from defense contracts if they are deemed to pose a threat to the integrity of the military’s supply chain. Historically, such designations have been reserved for companies with significant foreign influence or those suspected of being subject to the control of adversarial nations, such as telecommunications firms with ties to the Chinese government.
In this instance, the designation appears to be a novel application of the law, targeting a domestic company headquartered in San Francisco. A supply-chain-risk designation is intended to protect sensitive military systems and data from potential compromise, including risks related to foreign ownership, control, or influence (FOCI). By applying this label to Anthropic, the Pentagon is signaling that the startup’s refusal to grant unrestricted access to its models constitutes a security vulnerability that could impede military readiness or operational flexibility.
A Timeline of the Escalating Dispute
The breakdown in relations between the Pentagon and Anthropic did not occur in a vacuum. The conflict has been building for several months as the Department of Defense seeks to integrate generative AI into its "Replicator" initiative and other advanced programs.
In early 2024, Anthropic began high-level discussions with the Pentagon regarding the deployment of its Claude 3 and Claude 3.5 models within classified military environments. Unlike some of its competitors, Anthropic is structured as a Public Benefit Corporation (PBC) and has integrated a "Constitutional AI" framework into its development process, which prioritizes safety and alignment with specific ethical principles.
By mid-year, negotiations reportedly reached a stalemate. Anthropic representatives insisted on contractual language that would explicitly prohibit the use of their software for domestic surveillance programs targeting American citizens and for the development of fully autonomous weapons systems that operate without human intervention. The Pentagon rejected these stipulations, arguing that "all lawful uses" must be on the table to ensure the U.S. maintains a technological advantage over global adversaries like Russia and China.
Earlier this week, the tension became public when Anthropic published a blog post outlining its stance. The company argued that its refusal to support certain military applications was a matter of safety and corporate responsibility. On Friday morning, Secretary Hegseth responded with the "supply-chain risk" designation. By Friday evening, Anthropic issued a second statement, vowing to challenge the designation in federal court and claiming it had received no direct communication from the White House or the Department of Defense prior to the public announcement.
Reaction from the Technology Sector and Policy Experts
The Pentagon’s decision has been met with widespread condemnation from industry leaders and policy experts, many of whom view the move as an unprecedented overreach of executive power. Dean Ball, a senior fellow at the Foundation for American Innovation and a former senior policy adviser for AI at the White House, described the action as "the most shocking, damaging, and overreaching thing I have ever seen the United States government do." Ball emphasized that the designation effectively sanctions an American company for attempting to negotiate the terms of its service, suggesting it could drive innovation away from the United States.
The sentiment was echoed by Paul Graham, the founder of the influential startup accelerator Y Combinator, who characterized the administration’s behavior as "impulsive and vindictive." Within the AI research community, there are concerns that this move will create a chilling effect, discouraging other high-growth startups from engaging with the government. Boaz Barak, a researcher at OpenAI, stated that "kneecapping one of our leading AI companies is right about the worst own goal we can do," expressing hope that the decision would be reversed by "cooler heads."
The Divergent Path of OpenAI
As Anthropic faced a de facto ban, its primary competitor, OpenAI, appeared to take a different approach. On Friday night, OpenAI CEO Sam Altman announced that his company had reached a comprehensive agreement with the Department of Defense to deploy its AI models in classified environments.
Altman’s statement seemed to address the very issues that led to Anthropic’s exclusion. He noted that OpenAI’s agreement includes prohibitions on domestic mass surveillance and maintains the principle of human responsibility for the use of force, including in autonomous systems. According to Altman, the Department of Defense (referred to in his post as the Department of War, a term sometimes used rhetorically by industry figures to emphasize the gravity of the partnership) agreed that these principles are already reflected in existing law and policy.
The contrast between the two outcomes has led to speculation regarding the specifics of the negotiations. While OpenAI claims to have successfully codified its safety principles into a deal, Anthropic’s inability to do so suggests either a more rigid set of demands from the Pentagon in Anthropic’s case or a more uncompromising stance from Anthropic’s leadership regarding the legal language of the contract.
Legal Ambiguity and the Impact on Defense Contractors
The immediate practical consequences of the designation remain unclear, leading to confusion among hundreds of defense contractors. Many large-scale system integrators, such as Lockheed Martin, Raytheon, and Northrop Grumman, utilize a variety of AI tools in their internal workflows and within the products they deliver to the military. If these companies use Anthropic’s Claude for coding assistance, document analysis, or data processing, they may now be in violation of their federal contracts.
Legal experts have pointed out significant ambiguities in Secretary Hegseth’s directive. Alex Major, a partner at the law firm McCarter & English specializing in government contracts, noted that the Secretary’s announcement does not appear to be grounded in clearly defined statutory processes. "It is not mired in any law we can divine right now," Major said, referring to the sweeping nature of the ban on "any commercial activity" with Anthropic for anyone doing business with the military.
Under 10 U.S.C. § 3252, the authority to restrict vendors is generally applied to specific procurement actions. The claim that the Secretary can prevent a private contractor from using a specific software for its non-military clients, or for its own internal business, is a legal theory that will likely be tested in the coming weeks. Anthropic’s legal team has already argued that the Secretary lacks the statutory authority to enforce such a broad prohibition.
Broader Implications for National Security and the AI Industry
The designation of Anthropic as a supply-chain risk carries profound implications for the future of the U.S. national security innovation base. For years, the Department of Defense has struggled to bridge the "Valley of Death"—the gap between successful technology prototypes and large-scale military adoption. By blacklisting a major domestic AI provider, the Pentagon risks alienating the very ecosystem it has sought to cultivate.
Furthermore, there is the question of technological sovereignty. Anthropic is a multi-billion-dollar enterprise backed by major American corporations including Amazon and Google. It is a cornerstone of the domestic AI industry. Critics argue that by treating a domestic leader as a security threat, the government is inadvertently aiding foreign competitors. If American firms are restricted from using top-tier domestic models, they may be forced to rely on less capable alternatives or navigate a fragmented regulatory landscape that slows development.
The economic impact is also substantial. Anthropic has raised billions of dollars in venture capital and has become a preferred provider for enterprise-grade AI due to its focus on "Claude," which many users find more steerable and less prone to "hallucinations" than other models. A total ban on military-adjacent commercial activity could significantly impact Anthropic’s revenue and valuation, potentially leading to job losses and a reduction in R&D spending within the U.S. tech sector.
Conclusion and Outlook
The standoff between Anthropic and the Department of Defense represents a pivotal moment in the relationship between the state and the technology sector. It highlights the growing friction between the ethical considerations of AI developers and the operational requirements of a military seeking to maintain its global standing.
As Anthropic prepares its legal challenge, the defense industry remains in a state of flux. Contractors are seeking urgent clarification from the Pentagon’s Office of Acquisition and Sustainment regarding the implementation of the ban. Meanwhile, the executive branch’s use of supply-chain risk designations for domestic political or ethical disputes will likely face intense scrutiny from Congress and the judiciary.
The outcome of this dispute will likely set a precedent for how other emerging technologies—from quantum computing to synthetic biology—are regulated and integrated into the national security apparatus. For now, the "supply-chain risk" label remains a potent and controversial tool in the Pentagon’s arsenal, one that has transformed a contractual disagreement into a high-stakes battle over the future of American innovation.
About the Author
