Lifting export restrictions on Claude Mythos 5: the path to Fable 5 is open, but not for everyone
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 system for over 100 American organizations, including major corporations and government entities listed in Appendix A of the licensing list. Transferring the model no longer requires obtaining a separate license.
This move ends a two-week standoff between the Trump administration and Anthropic. Previously, exports of both models — Mythos and Fable 5 — were blocked after one of Anthropic's key investors, Amazon, expressed serious concerns about their safety. Company researchers warned that Fable 5 could be hacked for potentially dangerous purposes, which led to such stringent measures.
Mythos 5: From Secret Project to Broad Access
Until now, Mythos was used exclusively within the Glasswing program — a vulnerability discovery initiative involving about 150 organizations from more than 15 countries. Notably, this model identified critical errors in classified systems within hours during government testing, confirming its enormous potential in cybersecurity.
Now, the restrictions that prevented the launch of Mythos and Fable have been officially lifted. However, while Mythos gets the "green light" for a wide range of verified partners, the fate of Fable 5 remains uncertain. Unlike Mythos, access to Fable 5 was previously open to all subscribers, and for a time it was considered the most powerful public AI tool. Currently, its release to a broad audience is postponed, with no specific timeline announced.
The New Reality of AI Regulation
The situation surrounding Anthropic is just part of a global trend. On June 2, an executive order was signed, opening a voluntary channel for federal review of advanced AI models. Developers can submit their solutions for cybersecurity checks 30 days before launch. Simultaneously, Washington is tightening exports of AI chips to China, and now the same principles are being applied to access to the models themselves.
OpenAI followed the same path on Friday, restricting access to the most powerful version of GPT-5.6, called Sol, to about 20 partners who received government approval. Weaker versions — Terra and Luna — became available to the general public.
My expert opinion: We are witnessing the formation of a new control architecture in the AI field, where access to advanced models becomes a privilege, not a right. For the crypto industry, this means that the integration of powerful AI tools into DeFi and cybersecurity will occur with an eye on geopolitics. Fable 5 will likely receive approval, but with significant restrictions for non-U.S. jurisdictions. This sets a precedent that will change the landscape of AI development and use in the coming years.