Trump Bans Anthropic’s AI Model In Ongoing Attack

NEWS & RESEARCH

On June 12, 2026, the White House called Anthropic, an artificial intelligence (AI) start-up, to tell the company that it had less than 90 minutes to take down its latest AI models from the market. Government officials cited "national security concerns," without explaining what the specific problem was. Shortly thereafter, the Department of Commerce sent a written notice, ordering Anthropic to cut off access to foreign nationals—including visa holders working inside the US. Because there was no way to filter users by citizenship status in real time across dozens of cloud platforms, Anthropic had to pull the models entirely to comply. Anthropic had provided the models to the Commerce Department for pre-release testing, and the government initially didn’t raise any concerns about the product before abruptly demanding its removal. Sam Altman, the CEO of OpenAI, and Elon Musk, the owner of xAI, are both major allies of Trump, and their companies—which are some of Anthropic’s main competitors—have not faced similar government controls, despite having comparable capabilities. In March 2026, Anthropic balked at allowing the Trump administration to use its models for autonomous weapons systems or mass domestic surveillance.

SOURCES: New York Times | CNN | The Atlantic | Wall Street Journal | Foreign Policy | The Verge | Vox

ANALYSIS & OPINION

The Department of Commerce argued that letting foreign nationals access Anthropic's models constitutes an "export" under federal law, and that this classification gave the department the authority to regulate it. Such reasoning, though, contradicts the department's own prior guidance concluding that cloud-based access does not meet the legal definition of an export. Many national security experts believe that the administration's move was ideologically motivated, as Trump has described Anthropic as an "out-of-control Radical Left company." But by pulling a commercially deployed AI model from the market with 90 minutes' notice—no published rule, no congressional authorization, no transparent process—the administration effectively made itself the gatekeeper of which AI products can exist and for whom. Cybersecurity experts, industry officials, and even administration insiders have warned that if the administration can revoke their access to AI software on a whim, foreign businesses will turn to non-American alternatives, such as Chinese models.

SOURCES: Politico | The Economist | CNBC | The Atlantic | New York Times | Vox | TechCrunch

HOW TO FIX IT

Federal action:

• Consider the discussion draft of the bipartisan Great American AI Act (GAAIA). Many people, including the CEO of Anthropic, believe that artificial intelligence should be subject to transparent federal regulation with clear standards and processes—rules that would ensure safety while also preventing the executive branch from acting with unchecked authority. The GAAIA would be a meaningful first step in that direction.

Litigation:

• On June 23, 2026, less than two weeks after the initial directive, Legion—a U.S.-based legal-tech startup—filed a lawsuit in federal court challenging the export control directive, arguing that losing access to the AI models caused "immediate, irreparable and existential" harm to its business. The case is the first known legal challenge from a direct customer and may force a court to rule on whether the government had any legal authority to restrict access to a cloud-based AI model in the first place.

• Anthropic could file suit against the government over the recent directive on several grounds: that the Commerce Department acted arbitrarily and capriciously under the Administrative Procedure Act (APA) by creating new policy without following the normal rulemaking process; that the directive contradicts the agency's own prior guidance concluding that cloud access does not constitute an "export"; that it constitutes First Amendment retaliation, given the administration's documented hostility toward Anthropic; and that the executive branch lacks the unilateral authority to remove a commercial product from the market without congressional authorization.

• Anthropic is also in active litigation against the Department of Defense. When Anthropic refused to grant the military unfettered access to its models for use in warfare, the Pentagon responded by labeling the company a "supply chain risk"—barring defense contractors from using its tools. Anthropic sued, arguing the designation was unlawful retaliation. While that case concerns a different dispute, it raises the same core question: whether the government is using its power to punish Anthropic for being one of the few AI companies failing to “genuflect" before the Trump administration.