Anthropic export controls trigger debate over AI sovereignty

A US export control directive briefly forced Anthropic to suspend access to its Claude Fable 5 and Claude Mythos 5 models for users worldwide, prompting a wider debate in Europe and Canada over dependence on foreigncontrolled AI systems. The company said it received the order on June 12, 2026, after the models had only recently been made available. Anthropic said it had to disable access for all users because it could not separate foreign nationals quickly enough to comply with the directive. US officials said the action followed concerns about a jailbreak technique tied to Fable 5, while Anthropic argued the issue was limited and comparable to capabilities found in other frontier models. The company’s broader conflict with the US government had already escalated after it objected to military and surveillance uses for its technology. The episode intensified calls for AI sovereignty in Europe and Canada, where officials warned that critical AI tools can be switched off by another country’s decision. Regulators and ministers in several countries said the case highlights the risks of relying on a small number of US providers for advanced AI services.