Anthropic has received export approval for Mythos 5: over 100 US companies will gain access — Fable 5 is on the way
On Friday, the U.S. Department of Commerce officially lifted the export ban on Anthropic's Claude Mythos 5 model. This decision opens access to the advanced AI model for over a hundred American organizations, including major corporations and government entities listed in the so-called "Appendix A List."
This concludes a two-week standoff between the Trump administration and Anthropic, which arose from security concerns about Mythos and its more powerful counterpart, Fable 5. While Mythos is back in operation, Fable 5 remains unavailable to the general public for now, with its fate being decided at a higher level.
Mythos 5: From Glasswing to Full Approval
Commerce Secretary Howard Lutnick confirmed in a letter to Anthropic's Chief Technology Officer Tom Brown that a special license is no longer required to transfer Mythos 5 to organizations in "Appendix A." According to him, existing safeguards allow trusted partners to access the model.
Previously, Mythos was used exclusively within the Glasswing project, a vulnerability discovery program involving about 150 organizations from over 15 countries. Within hours of testing, the model identified critical errors in classified government systems, sparking interest in its broader deployment.
Notably, the initial block was triggered by concerns from Amazon—one of Anthropic's largest investors—regarding the models' safety. Company researchers warned that Fable 5 could be hacked for potentially dangerous purposes, leading to the temporary shutdown of both products.
Fable 5 Awaits—A New AI Regulation Regime Takes Shape
The situation with Fable 5 is more complex. Unlike Mythos, access to this model was previously open to all subscribers, and for a time, it remained the most powerful public AI tool. Now, its release has been postponed, with no specific timeline announced.
On June 2, an executive order was signed creating a voluntary federal review channel for advanced AI models. Developers can submit their solutions for cybersecurity checks 30 days before launch. This effectively establishes a new control mechanism that Washington is actively implementing, while simultaneously tightening AI chip exports to China.
Interestingly, OpenAI has followed a similar path: on Friday, the company restricted access to the most powerful version of GPT-5.6, named Sol, providing it only to about 20 partners who received state approval. The weaker versions, Terra and Luna, became available to the general public.
Initially, the blocks were driven by fears that Chinese entities could acquire the technology. Attention focused on South Korean operator SK Telecom, which was added to the Glasswing list in early June and then had its access restricted. SK Telecom denied any connection to China.
Leaders of major cybersecurity companies, including former Facebook security chief Alex Stamos, as well as representatives from Nvidia, Adobe, and Zoom, urged authorities to abandon restrictive measures. Now that Mythos 5 has received the "green light," all eyes are on Fable 5—whether it will receive similar approval in the coming days.
Expert opinion: The lifting of the ban on Mythos 5 is not just a victory for Anthropic, but a signal of a new AI control architecture taking shape. Washington is setting a precedent where access to the most powerful models will be strictly regulated, and companies will be forced to undergo voluntary, yet essentially mandatory, review. Fable 5 will serve as a stress test for this system.