Anthropic halts access to top AI models after U.S. export order

Anthropic abruptly cut off access to its two most advanced AI models after a U.S. export control directive ordered the company to suspend access for foreign nationals on national-security grounds. The company said the order arrived Friday afternoon, gave no specific details, and left Anthropic with no practical option but to disable Fable 5 and Mythos 5 for customers around the world.

Anthropic said it believes the government may have acted because it identified a way to bypass, or “jailbreak,” Fable 5’s safeguards. The company called the episode a “misunderstanding” and said it wants to restore access as soon as possible. Other Anthropic models were not affected. TechCrunch said the directive reached the company at 5:21 p.m. ET.

The move landed just days after Anthropic publicly launched Fable 5 on June 9, 2026, pitching it as a guarded public version of the same underlying model as Mythos 5. Anthropic said Fable 5 was built for enterprise customers and paid subscribers, with safety classifiers designed to route high-risk prompts away from the model. Mythos 5, by contrast, had been reserved for a small group of vetted cyber defenders and infrastructure providers, first through Project Glasswing, a collaboration with the U.S. government. Anthropic said the two models were priced at $10 per million input tokens and $50 per million output tokens, less than half the price of Claude Mythos Preview.

Anthropic has said Mythos 5 is its strongest cybersecurity model and can work autonomously for longer than any previous Claude model. The company said Fable 5 also ranked as the most capable AI model available to the public on benchmark tests, with new safety controls intended to block or reroute high-risk requests involving cybersecurity and biology. The sudden shutdown now separates commercial release plans from government security concerns, and it does so at the moment the product line was meant to showcase safer frontier AI.

The broader significance reaches well beyond Anthropic. The export controls mark the most significant U.S. step yet to restrict access to advanced AI models, and they came about 10 days after President Donald Trump signed an executive order creating a federal framework to vet national-security risks in advanced AI systems before public release. That framework was voluntary, but the Anthropic case shows how quickly a government order can redraw who gets access to frontier technology, where the line is drawn between domestic and foreign users, and how model access itself is becoming a tool of geopolitics. The Commerce Department did not immediately comment.