Anthropic forced to disable Fable 5 & Mythos 5 over US export controls on foreign nationals

The Trump administration issued export-control directives forcing Anthropic to suspend access to its two most powerful AI models, Fable 5 and Mythos 5, for any foreign national—both inside and outside the United States, including Anthropic's own foreign-born employees. Citing national-security concerns, the Commerce Department invoked export-control authority over models that exhibit cybersecurity capabilities and potential vulnerabilities to jailbreaking. Because Anthropic cannot reliably sort users by nationality on a shared API, the company disabled both models globally rather than attempt partial enforcement.

The government's stated concern is a narrow jailbreak of Fable 5 that could enable unauthorized access to security-vulnerability information. Anthropic disputes both the severity and the need for such a drastic measure, saying the capability is narrow and widely available in other models, and arguing the blanket ban is disproportionate. The disagreement signals a deeper tension: the directive represents the first export-control measure targeting specific AI models rather than silicon or hardware, and it functions as a de facto kill switch on frontier capability access for non-US users.

For global organizations and AI architects, this marks a watershed moment in AI governance: US frontier models—Anthropic and OpenAI—are now subject to nationality-based access restrictions with minimal notice and opaque security criteria. European policymakers, including France and the UK, cited the move as validation for developing sovereign AI capabilities and reducing dependence on US systems. The action also highlights operational risks for teams with foreign-born talent and multinational users relying on US AI services.