The United States has spent years restricting the export of advanced AI chips. Now it is moving beyond hardware and targeting the AI models themselves. In a dramatic step that could reshape the global AI race, the Trump administration has ordered Anthropic to block foreign access to its most advanced systems, citing national security concerns.
The Trump administration has ordered AI company Anthropic to prevent foreign governments, companies and individuals from accessing its most advanced artificial intelligence models, marking one of the most significant expansions of US export controls into the AI software sector.
According to Reuters and Axios, Commerce Secretary Howard Lutnick informed Anthropic that its flagship models, Mythos 5 and Fable 5, would be subject to export restrictions both outside the United States and for foreign nationals within the country.
The move forced Anthropic into an extraordinary response.
Rather than attempting to separate American and foreign users immediately, the company announced it would disable access to the models entirely while it works to comply with the directive. Anthropic said the order effectively required it to suspend access for all customers.
At the center of the dispute is a reported security concern involving Fable 5.
Government officials believe there may be a way to “jailbreak” the model’s safeguards and use it to identify software vulnerabilities. Anthropic says it was provided only verbal evidence of what it described as a narrow and limited vulnerability and argued that such concerns do not justify recalling a commercial AI model.
The fight is no longer about who can build the most powerful AI. It is increasingly becoming about who is allowed to use it.
The decision represents a major shift in US AI policy.
For years, Washington’s efforts focused largely on restricting advanced semiconductors and chipmaking equipment from reaching rivals such as China. This latest action instead targets direct access to frontier AI systems themselves, potentially creating a new framework for controlling the spread of advanced artificial intelligence capabilities.
The order also affects close US allies.
Reports indicate the restrictions apply broadly to foreign nationals regardless of location, meaning users in countries such as the United Kingdom, Canada, Australia and others could lose access despite being long-standing American partners.
Anthropic has pushed back strongly against the government’s decision.
The company said it believes there has been a misunderstanding and warned that applying the same standard across the AI industry could effectively halt the deployment of future frontier models. Executives have expressed concern that regulators are reacting to hypothetical risks without sufficient evidence of actual harm.
The confrontation comes against the backdrop of a complicated relationship between Anthropic and the Trump administration.
Earlier this year, tensions emerged after disputes involving military applications of AI and government concerns about national security. While relations appeared to be improving in recent months, the latest order has reopened a major conflict between the company and federal officials.
Amazon reportedly played a role in bringing the issue to government attention.
Reuters reported that Amazon CEO Andy Jassy raised concerns with senior administration officials after information surfaced about a potential method for bypassing safeguards in Anthropic’s latest systems. Those concerns contributed to the administration’s decision to intervene.
The broader implications could be enormous.
If the restrictions remain in place, they may establish a precedent allowing governments to treat advanced AI models as strategic national assets similar to sensitive military technologies. Other AI companies could eventually face similar export controls if their systems reach certain capability thresholds.
For now, Anthropic says it is working with federal officials to resolve the issue and restore access.
But the episode has already sent a clear message across the technology industry: the next battleground in the global AI race may not be building powerful models, but controlling who gets to use them.
