Anthropic announced on its official X account on July 1 that the U.S. Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5, and the two flagship models will restore user access starting July 2. On June 12, the Bureau of Industry and Security (BIS) issued an order citing national security, suspending access to the two models for "any foreign person."
Anthropic Official Statement and July 2 Access Restoration Timeline
Anthropic's official X account issued a statement on the morning of July 1: "We have been notified that the U.S. Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5. We will restore access starting tomorrow and will share updates soon. We appreciate our users' patience and everyone who worked with us to redeploy the models."
According to the statement, the access restoration date is July 2, 2026; Anthropic said it will release detailed updates separately, and specific restoration details had not been announced as of the announcement.
Triggers for the Export Control Order: Jailbreak Risks and South Korean Telecom Incident
According to public reports, the trigger background for the June 12 BIS order includes two parallel events:
Jailbreak Risk Warning: Partners such as Amazon warned relevant agencies that Claude Fable 5 and Mythos 5 have security risks of being "jailbroken."
Mythos 5 Overseas Release Incident: Mythos 5 was reportedly released to a South Korean telecom operator with ties to China, triggering national security concerns.
Regulatory Precedent Significance: AI Models Listed as Controlled Export Items for the First Time
The June 12 BIS order is the first time the U.S. government has used export controls to restrict access to domestic frontier AI models, with regulatory logic similar to the control framework for advanced semiconductors (such as H20 chips). Because Anthropic could not reliably filter accounts by user nationality, the order effectively forced a global 18-day shutdown of the two models for all users, not just foreign users.
This case formally establishes a regulatory precedent that AI models can be controlled in terms of access scope via export controls, with the regulatory scope expanding faster than the industry expected.
Frequently Asked Questions
What follow-up steps does Anthropic need to take after the controls are lifted?
Anthropic's statement said it will "share updates soon" and will begin restoring access on July 2. Specific restoration procedures, phased rollout order, and complete recovery timeline had not been announced as of the July 1 announcement.
Which users were affected during the export control period?
The BIS order required suspending access for "any foreign person." Anthropic could not reliably filter accounts by nationality and was forced to shut down both models entirely, affecting all users including those within the United States, with a shutdown period of 18 days.
What precedent does this case set for the compliance obligations of future AI companies?
The BIS order on Anthropic's flagship models is the first known case of classifying frontier AI models as controlled export items, extending the regulatory logic from advanced chips to model access itself. This case implies that AI companies may need to assess their models' nationality filtering capabilities and control mechanisms for foreign user access.
