Why the U.S. Government Ordered Anthropic to Shut Down Fable 5 and Mythos 5 for the Entire World

At 5:21 p.m. ET on June 12, 2026, the U.S. Commerce Department's Bureau of Industry and Security (BIS) issued an export control directive that did something no government had done before: it ordered a company to pull a specific, commercially deployed AI model away from the entire world. The target was Anthropic's Fable 5 and Mythos 5, the most advanced models from the maker of Claude .

The directive banned access for any foreign national, whether inside or outside the United States, including Anthropic's own non-U.S.-citizen employees . Because the company could not reliably segment its user base by nationality in real time, it was forced to disable both models for all customers globally. The result was an instant, worldwide blackout of Anthropic's most capable AI systems .

What the export control directive actually said

The order, issued under national security authorities, required Anthropic to "suspend all access to Fable 5 and Mythos 5 by any foreign national" . Commerce Secretary Howard Lutnick communicated the directive directly to Anthropic CEO Dario Amodei via a letter the same day .

The scope was sweeping. It covered not just people outside the U.S., but any foreign national physically present inside the country, including Anthropic's own foreign-national staff . Because real-time nationality-based access controls didn't exist reliably, the company concluded it had no choice but a total takedown .

The government's stated concern was a potential method to bypass, or "jailbreak," the models' built-in safety guardrails . The BIS provided no formal written evidence, only what Anthropic described as verbal notice .

Why this was a first-of-its-kind action

Before June 12, U.S. export controls on AI had focused on the hardware layer—most notably, restricting sales of advanced Nvidia chips to China . This directive was different. It was the first time the government applied an export-control instrument directly to a commercial AI model rather than the underlying compute .

Analysts noted the move effectively converted export-control law into a live access-control lever over deployed software, reshaping the relationship between frontier AI labs and the state in a way no existing governance framework anticipated . As one commentator put it, frontier models were no longer just a cloud service; they had become a controlled dual-use technology .

Anthropic's response: compliance with a sharp pushback

Anthropic complied with the legal directive immediately but did not stay quiet. In a public statement posted the same evening, the company announced it would "abruptly disable Fable 5 and Mythos 5 for all our customers" to ensure compliance . Access to all other Claude models remained unaffected .

The company's pushback was direct and public. It said the government had given only "verbal evidence of a potential narrow, non-universal jailbreak" and called the entire directive a "misunderstanding" of the models' capabilities .

Anthropic argued the jailbreak in question was narrow, non-universal, and produced results already achievable on other publicly available models including GPT-5.5 . It also raised a values-based objection: a blanket exclusion of all non-U.S. citizens, the company said, was not compatible with its principles .

A more practical problem was logistics. The BIS order required nationality verification on all API calls, a capability Anthropic said didn't exist in any reliable form . With no way to comply selectively, the company pulled the plug for everyone.

European Commission reaction: discrimination and sovereignty

The European Commission responded within 48 hours on two fronts that have long defined transatlantic tech tensions.

"Should not be discriminatory"

EU spokesperson Thomas Regnier stated plainly that U.S. export controls on Anthropic "should not be discriminatory" against EU partners . The Americans-only order had locked European governments, corporations, and institutions out of frontier AI models with no warning and no transition period .

The Commission launched a formal probe into the U.S. directive and its impact on European customers, while continuing discussions with technology partners about the cybersecurity implications of advanced models .

A sovereignty wake-up call

European politicians used sharper language. The episode was widely called a "wake-up call" for Europe's need for technological sovereignty . Regnier himself said the blockade further reinforced "Europe's need for technological sovereignty" .

The logic was blunt: if the U.S. government can unilaterally cut off European access to the most capable AI models with a Friday afternoon letter, the EU cannot afford to depend on American companies for its AI infrastructure .

This was not entirely new territory. Only a month earlier, the EU had gained access to Mythos after weeks of negotiations, and tensions had already surfaced when Anthropic initially withheld the model from the bloc .

Broader implications

The June 12 directive established a precedent whose consequences are still unfolding. For the first time, a government demonstrated it can and will use export-control authority to kill access to a specific deployed AI model, not for a foreign adversary but globally, affecting allies as collateral .

For AI companies, the episode introduced a new operational risk: a national security letter can arrive at 5:21 p.m. and require a product to disappear by morning with no gradual phase-out, no customer warning, and no appeal window .

For governments outside the U.S., the message landed hard. If frontier AI access depends on the goodwill—or the internal security calculus—of a single country's executive branch, the case for sovereign AI capacity moves from abstract policy paper to urgent priority.