Anthropic's Fable Classed as Munition: US Blocks AI Model Worldwide

On June 9, Anthropic released the generative model Fable (together with Mythos 5), one of the most powerful in its lineup. Just three days later, the US government declared it a military munition and barred foreign users from accessing it. The result: the model became unavailable to the entire world.

Fable Under Export Controls

The US Department of Commerce applied export control law: Fable and Mythos 5 were classified as "dangerous munitions," with access banned for any foreign national. This is the first time an American commercial AI model has received such a status, on par with traditional military technology.

Anthopic found itself in a dead end: technically, the company cannot reliably separate American and foreign users. Identifying citizenship across millions of accounts is a task most platforms have no infrastructure for. IP addresses are masked by VPNs, and payment data often does not match citizenship. As a result, the company had to shut off access to both models entirely, for everyone, including US citizens.

Users quickly launched a petition, FreeFable, demanding the block be lifted.

"The government's actions will not help.

The problem is not a specific model but the general trend of growing AI capabilities," said Bruce Schneier, security specialist, Harvard Kennedy School.

Why the Ban Will Not Help

Classifying Fable as a munition is a political decision, not a technical barrier. Cybersecurity expert Bruce Schneier and researcher Nathan Sanders explain in a column for The Guardian why controlling individual models does not solve the problem:

• AI capabilities will keep growing regardless of the fate of one model or company
• Export controls create an illusion of security without curbing the global spread of knowledge
• Other countries and independent researchers are building similar systems without American restrictions
• A block for everyone does real economic damage without removing the threats
• Mechanisms for international coordination to regulate AI at this level do not yet exist

The authors note that a real solution would require collective global action, and achieving that kind of coordination in the current geopolitical climate is not yet possible.

A Technology That Cannot Be Shut Away

The Pandora's box metaphor in the column's headline is apt: Fable was not a point of no return, but it marked a turning point. AI has reached a level at which states are beginning to see it as both a strategic asset and a threat to national security.

Schneier's key argument is that the problem is systemic. One company and one model can be banned, but the entire trajectory of the technology cannot be stopped. Fundamental AI research is open: algorithms are published in peer-reviewed papers, training data is available to a wide range of players, and training methods are not state secrets.

The American block is the first of its kind, but unlikely to be the last. As the next generations of models grow more capable, regulatory pressure will mount. The question stops being "to ban or not to ban" and becomes "how to build a system of governance for a technology that cannot be fully stopped."

What This Means

The Fable precedent moves the AI debate from the technological arena into the geopolitical one. For developers, businesses and users, it is a signal: access to frontier models is now determined not only by corporate decisions but also by the political will of regulators.

Companies will have to build geolocation and verification mechanisms into their product architecture in advance, or risk the next ban catching them as off guard as it did Anthropic.

There is still no unified international response to this challenge, and models keep getting more powerful.