Developing: Trump Orders Federal Ban on Anthropic AI After Company Refuses Pentagon Demand for Unfettered Access

At the heart of the dispute is a fundamental question about who controls how AI tools are deployed in military and national security contexts. The Pentagon demanded what it described as "any lawful use" of Anthropic's AI systems — essentially unfettered access without the company's safety guardrails or usage restrictions.

Anthropic CEO Dario Amodei said he would rather terminate the company's government contracts than comply. His specific concerns centered on two applications: mass domestic surveillance and fully autonomous weapons systems. Amodei framed the issue as a matter of responsible AI deployment rather than a political stance, stating that these represented hard boundaries the company would not cross regardless of the client.

Anthropic has been providing AI tools to the US government and military since 2024, through a contract facilitated in part by defense technology firm Palantir. The company said that if the Department of Defense chose to stop using its tools, it would "work to enable a smooth transition to another provider."

Defense Secretary Hegseth summoned Amodei to Washington for a meeting earlier in the week that ended with two ultimatums: grant the Pentagon unrestricted access, or face the consequences. The Pentagon threatened to invoke the Defense Production Act — which allows the government to compel companies to produce goods for national defense — and to label Anthropic a "supply chain risk," which would effectively bar any company doing business with the military from also working with Anthropic.

In a remarkable display of industry solidarity, OpenAI CEO Sam Altman publicly backed his direct competitor. In an internal memo seen by the BBC, Altman told staff that OpenAI shared the same "red lines" and that any of its defense contracts would also reject uses that were "unlawful or unsuited to cloud deployments, such as domestic surveillance and autonomous offensive weapons."

The support from Altman is particularly notable given the history between the two companies. Amodei and several other OpenAI employees left to found Anthropic after disagreements with Altman over AI safety priorities. The two startups now compete directly for consumers and enterprise customers. Yet on this issue, the industry appears united.

"I do not fully understand how things got here," Altman wrote. "But regardless of how we got here, this is no longer just an issue between Anthropic and the DoW; this is an issue for the whole industry and it is important to clarify our stance."

The response extended beyond executive suites. Groups representing roughly 700,000 tech workers at Amazon, Google, and Microsoft — all companies with their own Pentagon contracts — signed an open letter urging their employers to "refuse to comply" with the Pentagon's demands. The Alphabet Workers Union issued a separate statement declaring: "Tech workers are united in our stance that our employers should not be in the business of war."

The breadth of the response suggests the Trump administration's approach to Anthropic has galvanized resistance across the technology sector in ways that individual policy disputes rarely have.

The government's legal authority to compel Anthropic's compliance is a matter of active debate. A former Department of Defense official, speaking anonymously, told the BBC that the basis for threatening Anthropic with either the Defense Production Act or the supply-chain risk designation was "extremely flimsy."

The Defense Production Act has historically been used to compel manufacturing of physical goods — most notably during the COVID-19 pandemic to increase production of medical supplies. Applying it to software services and AI model access would represent a significant expansion of the statute's scope and would almost certainly face legal challenge.

The supply-chain risk designation is equally novel in this context. The designation has traditionally been applied to foreign companies — most notably Chinese telecommunications firms like Huawei — that are deemed national security threats. Applying it to a domestic AI company valued at $380 billion would be unprecedented and could have chilling effects on technology companies' willingness to contract with the federal government.

Former Air Force Secretary Frank Kendall called Trump's order "outrageous," adding to a growing chorus of former defense officials who view the administration's approach as counterproductive to national security interests.

The confrontation between the Trump administration and Anthropic raises broader questions about the AI industry's relationship with government contracts and defense spending. Several dimensions of the dispute carry market significance.

For Anthropic specifically, the $200 million Pentagon contract represents a small fraction of the company's revenue and its $380 billion valuation. As the former DoD official noted, the situation is "great PR for them and they simply do not need the money." However, the supply-chain risk designation — if actually implemented — could create complications for Anthropic's enterprise customers who also do business with the military.

For publicly traded AI-adjacent companies, the implications are more complex. Microsoft (MSFT), Amazon (AMZN), and Alphabet (GOOG) all hold significant defense contracts and also operate cloud infrastructure that hosts AI models, including Anthropic's. If the supply-chain risk designation extends to companies that merely host or integrate Anthropic's technology, the ripple effects could be substantial.

The broader concern for the AI sector is whether the dispute signals a willingness by the current administration to use government contracting as a lever to override companies' safety and ethical guidelines. If AI companies face a binary choice between unlimited government access and losing federal contracts, some may choose to avoid government work entirely — which could ironically set back the military's AI capabilities relative to adversaries.

Defense-focused technology firms like Palantir (PLTR), which facilitated Anthropic's original Pentagon contract, may also see their role as intermediaries complicated by the precedent.

The Trump-Anthropic confrontation crystallizes a tension that has been building across the AI industry for years: who gets to decide what limits exist on the deployment of increasingly powerful AI systems?

Supporters of the administration's position argue that the federal government — and particularly the military — must have access to the most capable AI tools available without company-imposed restrictions. In this view, allowing private companies to dictate the terms of national defense creates an unacceptable dependency on entities that may not share the government's priorities.

Critics of the administration's approach, including many AI researchers and the tech workers who signed the open letter, argue that the specific applications at issue — mass surveillance and autonomous weapons — represent widely recognized red lines that exist for sound ethical and practical reasons. They point to international norms around autonomous weapons and the potential for AI-enabled surveillance to undermine civil liberties domestically.

Anthropic's stance places it in a unique position in the AI industry. The company was founded specifically around AI safety principles, and its refusal to compromise those principles — even in the face of presidential threats — has earned it support from unexpected quarters, including its primary competitor. Whether that stance is sustainable in the face of escalating government pressure remains an open question.