Anthropic Blocks Fable 5 and Mythos 5 Due to Trump Export Control Order

Anthropic has spent several tense days attempting to restore operations of its latest models, Fable 5 and Mythos 5, after the Trump administration unexpectedly demanded that access be closed to all foreign citizens—including users physically located within the United States and the company's own employees.

What Happened and How Quickly

The government's requirement caught Anthropic off guard. The company was forced to immediately block Fable 5 and Mythos 5 for all users falling under the definition of "foreign citizen"—a definition that turned out to be far broader than initially anticipated. Unlike standard regulatory processes, which typically provide for a transition period, this order required immediate compliance.

A particular problem arose from the fact that the restrictions affected the company's own employees. A significant portion of Anthropic's researchers are specialists who came from other countries but have worked in the United States for years. For them, losing access to their own working tools meant an immediate operational crisis. The company had to simultaneously address legal issues and technically manage access in real time.

In an official statement, Anthropic reported that the government cited "national security authorities" as the basis for "export controls." No public explanation was provided as to which specific law was being applied or why. This placed the company, its lawyers, and industry observers in an extremely awkward position: the requirement had to be complied with, but its legal nature could not be understood.

An Unprecedented Precedent

Experts in technology policy are unanimous: nothing like this has happened before.

"As far as I know, this is the first time that U.S. export controls

have been used to restrict access to an AI model in this manner."

Until now, American export controls in high-tech industries have primarily concerned chips and specialized equipment—for example, high-performance GPUs, whose importation is restricted for a number of countries. Applying this mechanism to cloud AI services represents a fundamentally different step: what comes under control is not a physical commodity, but the right to use a software model over the internet. This changes the very logic of how regulators can influence the AI industry.

Who Was Affected

The scope of restrictions proved to be far broader than the conventional understanding of "export controls" would suggest:

• Foreign citizens legally residing and working in the United States
• Anthropic employees with foreign passports
• International corporate clients using the Claude API
• Academic researchers from countries outside the "whitelist"
• Company partners with development teams abroad

If such orders become the norm, major AI laboratories will face a stark choice: restrict hiring of foreign specialists—or accept that employees could lose access to key company products at any moment.

What This Means

This episode marked a new line in American technology policy: export controls can extend not only to hardware, but also to software products—and can be introduced without public legal justification or advance notice. For AI companies, this represents a new type of regulatory risk for which the industry is not yet prepared. For the international market, it is a signal that access to cutting-edge American models could come under threat on grounds that were never announced in advance.