The US has lifted export controls on Anthropic AI models
The US government has lifted export controls on Anthropic’s latest AI models – Claude Fable 5 and Claude Mythos 5, enabling their availability in the international market once again.
Anthropic plans to gradually restore access to the Fable 5 AI to users worldwide starting from July 1 through platforms such as Claude.ai, Claude Platform, Claude Code, and Claude Cowork. By July 7, users on Pro, Max, Team, and some Enterprise plans will be able to access the model within 50% of their weekly limits. A shift to a paid credits system is planned later. Additionally, the company promises to restore access via AWS, Google Cloud, and Microsoft Foundry.
For the Claude Mythos 5 model, Anthropic has already granted access to a select list of organizations in the US. The company is working with the government to expand access for other domestic and international partners. Anthropic representatives expressed gratitude to users for their patience and to those who facilitated the redeployment of the models.
Prior to restoring access, on June 12, 2026, Anthropic was forced to deactivate Fable 5 and Mythos 5 models due to a US government export control directive. This required immediate access restriction to the products for all foreign nationals. The absence of a reliable tool to verify users’ nationalities in real-time prompted the company to temporarily halt access to the models.
Claude Fable 5 is the company’s most powerful model for general use and is characterized by high performance in programming, data handling, and document analysis. However, it declines to process certain queries in areas such as cybersecurity, biology, chemistry, and distillation. In such cases, the request is redirected to the Claude Opus 4.8 model. Mythos 5, in turn, is used for vulnerability detection in cybersecurity, and even by the US National Security Agency, despite Pentagon warnings about potential risks.
| Model | Key Characteristics | Current Status |
|---|---|---|
| Claude Fable 5 | High performance in programming, document analysis | Available from July 1, usage limitations |
| Claude Mythos 5 | Autonomous execution of complex tasks, cybersecurity | Access for selected organizations in the US |
