US export ban on Anthropic's Fable 5 blocks non-citizens even inside America

The US government imposed export controls on Anthropic's Fable 5 / Mythos 5 model on June 13, restricting access to US citizens only. The order took effect immediately, and Anthropic disabled all access while determining which users qualify under the new rules. Non-citizens inside the United States are also barred from using the model, a restriction that extends to Anthropic's own employees.

The government cited jailbreak exploits that bypass Anthropic's safety systems, though no specific examples were provided. Anthropic's public response stated that the company believes "the government should have the ability to block unsafe model deployments through a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not meet those principles." The company said it has a report showing GPT-5.5 is similarly vulnerable but remains unrestricted.

Anthropic promised a fuller response within 24 hours, planning to address the technical claims and the fairness of the enforcement process, which it argues lacks the transparency and technical grounding required for credible export control.

What stands out

1. 01Citizenship-based model access is unprecedented. The order blocks non-citizens from using Fable 5 / Mythos 5 even on US soil, marking the first AI export control tied to individual citizenship status rather than geography alone.
2. 02No disclosed vulnerability details. The government provided no public examples of the jailbreaks that triggered the ban, leaving practitioners and researchers unable to assess the technical justification.
3. 03Selective enforcement raises fairness questions. Anthropic claims GPT-5.5 has similar vulnerabilities but faces no restrictions, suggesting inconsistent application of the export control criteria.
4. 04