The Policy At The Center Of The Dispute

Anthropic entered into a reported $200 million defense contract last year to integrate its AI models, specifically the Claude family, into classified systems and military operations. In this initial agreement, the DoD accepted Anthropic’s acceptable use policy, including two “red lines” the company has been unwilling to compromise on: mass domestic surveillance and fully autonomous lethal weapon systems. The department previously described its collaboration with Anthropic as highly positive, selecting Claude over competing systems for defense applications.

Tensions emerged this January after Defense Secretary Pete Hegseth issued an artificial intelligence strategy memorandum directing procurement officials to incorporate “any lawful use” language into all AI-related defense contracts, meaning the department could use AI for any lawful use it sees fit. The stated objective was to remove any operational constraints that could limit the government’s ability to deploy AI capabilities in support of national security.

Anthropic resisted modifying its existing agreement due to concerns over the ambiguity of the “any lawful use” phrase, maintaining that two safeguards in Claude’s design — the red lines mentioned above — should remain intact.

The directive introduced a fundamental disagreement over whether a private technology vendor could impose operational restrictions on systems used in national defense. Negotiations quickly escalated into a Pentagon ultimatum requiring Anthropic to remove its AI guardrails for unrestricted military use or face consequences, including the agency terminating its contract and potentially designating the company a supply chain risk.

On the eve of the imposed deadline, Anthropic CEO Dario Amodei publicly stated the company “cannot in good conscience accede” to the Pentagon’s demands. Subsequently, the Pentagon and Executive Office officially labeled Anthropic a supply chain risk to national security, effectively barring the company from participating in all government-related technology ecosystems.

Following the designation, Anthropic filed a legal challenge arguing that the government’s action constitutes an overreach that could undermine the independence of commercial AI providers.

Industry Fallout

Among individual AI users, Anthropic’s stance has strengthened its reputation as a company prioritizing ethical AI practices and applications. However, the dispute is unlikely to influence enterprise adoption one way or another as long as large organizations continue to evaluate AI providers primarily on performance, reliability and integration capabilities for a given problem set rather than individual policy positions.

The designation triggered immediate legal action from Anthropic, and in a significant development, a federal judge has since blocked the supply chain risk designation with a temporary injunction. Judge Rita Lin cited concerns in her ruling that the Defense Department’s action may constitute retaliation against Anthropic for exercising its First Amendment rights. Although the lawsuit remains ongoing, this ruling represents an early setback for the Pentagon.

Several large tech companies, including Google, Amazon, Apple and Microsoft, have backed Anthropic, claiming that the Pentagon’s actions demonstrate a dangerous precedent of government intervention. Historically reserved as a defense against foreign adversaries, this is the first time in history that a U.S. company has been designated, or even threatened, with supply chain risk status. If ultimately upheld, such a designation could prevent Anthropic from engaging with any company currently associated with the U.S. government, which encompasses a majority of its business dealings. More broadly, this dispute addresses a shift in government perception of frontier AI models as they are increasingly treated as strategic national security infrastructure.

Another notable development in the dispute includes the involvement of OpenAI CEO Sam Altman. Prior to the Pentagon’s imposed deadline, Altman publicly expressed solidarity with Anthropic’s principles, stating that OpenAI shared the same red-line restrictions for AI systems. However, hours after Anthropic’s designation, OpenAI announced its own defense agreement for classified operations with the Pentagon.

The announcement sparked a largely negative reaction from the public and OpenAI employees alike, with some OpenAI employees electing to resign over the new Pentagon agreement. In a matter of days, the resulting decline in public opinion contributed to a 295% increase in ChatGPT uninstalls relative to a baseline of approximately 9%, while Claude installs surged by up to 51%, briefly reaching the No. 1 position on U.S. app stores, as reported by TechCrunch’s Sarah Perez.

Altman’s public support for Anthropic before finalizing OpenAI’s own defense agreement represented contradictory positions and may have permanently damaged public trust in the platform. From a business perspective, OpenAI is heavily dependent on government support and funding. Quickly securing an agreement to potentially establish itself as the go-to AI company for U.S. government operations creates an incredible opportunity for OpenAI, especially while pursuing costly data center construction.

Altman himself has responded to the backlash, stating that, despite being well-intentioned, the rushed nature of the agreement presents a perception that the action was opportunistic and sloppy. OpenAI has since amended its agreement to strongly affirm the company’s principles against domestic surveillance and autonomous lethal decisions. However, this does not restrict governmental authority to designate “any lawful use” in support of national security.

Soon after, several Google and OpenAI employees signed an amicus brief in favor of Anthropic to provide expert opinions and relevant information to aid in the court’s decision.

Possible Outcomes

The outcome of this legal battle could shape future contracts between the government and AI developers. If the court permanently overturns the designation, the case could become a landmark precedent for AI governance. Such a ruling could strengthen private AI developers to enforce ethical constraints on the use of their systems, even when dealing with national governments.

This outcome would herald Anthropic’s Claude as a moral and ethical AI solution, aiding in public perception, as the threat of AI overtaking job markets has sparked widespread disapproval of AI practices. More broadly, the outcome would heavily influence the structure of future contracts between governments and AI developers, potentially establishing clear boundaries between national security requirements and vendor-defined safety policies.

Alternatively, if the court lifts the block of the supply chain risk designation, Anthropic could face significant commercial restrictions. The company could lose access not only to direct federal contracts but also to all government-adjacent markets and partnerships, which would materially affect its growth prospects.

Despite public support, this outcome may force the company to find some way to address the government’s needs while still operating in a fashion that it finds acceptable or, if unsuccessful in doing so, potentially disband or diversify for survival. This would likely allow competitors such as OpenAI and Google to benefit from the resulting redistribution of talent, partnerships and investment.

This could also establish a precedent in which governments compel technology providers to modify product safeguards as a condition of participating in defense ecosystems, which Anthropic argues could gravely violate constitutional rights and safety.

The Bottom Line

The U.S. government has applied continuous pressure against Anthropic, a U.S. company, only to be met with increasing resistance. The Pentagon has asserted its position that the military, not a vendor, determines how technology is used during wartime operations. With competitors such as Google, Amazon and Meta supporting Anthropic, the court system is set to decide the appropriate nature of government intervention in the AI industry and the legal use of AI in national security measures. However, the recent court ruling blocking the designation indicates that the Pentagon’s authority may face meaningful constitutional limits.

In general, it is uncertain whether any single AI company will continue to be successful in the future. However, unlike OpenAI and Anthropic, companies with large, diversified revenue streams, such as Google and Microsoft, will continue to thrive in their existing AI market positions and deep integration within enterprise work environments.

Regardless of the final legal determination, the dispute showcases a structural tension in the AI industry: advanced AI systems are simultaneously commercial software products and strategic national security assets. As governments increasingly rely on frontier AI capabilities, conflicts between vendor-defined safety policies and federal operational authority could become a frequent issue if not thoroughly resolved. Consequently, the resolution of this case has the potential to permanently shape future contracts between governments and AI developers, influencing the future governance of artificial intelligence.