The U.S. administration has officially lifted the ban on exporting Anthropic's Claude Mythos 5 model. This decision, made on Friday, opens access to the advanced AI system for over 100 organizations, including major corporations and government agencies listed in the special Appendix A. A separate license is no longer required to transfer the model—compliance with established criteria is sufficient.

This move ends a two-week standoff between the Anthropic team and the presidential administration. Previously, both models—Mythos and Fable—were blocked due to concerns from investor Amazon, as well as warnings from the company's internal researchers about the potential for Fable 5 to be hacked for malicious purposes.

Mythos 5: From Closed Testing to Widespread Deployment

Before the restrictions were lifted, Mythos 5 was used exclusively within the Glasswing project—a vulnerability discovery program involving about 150 organizations from 15 countries. Notably, this model identified critical errors in classified systems within just a few hours during government testing. Now, leading U.S. companies and agencies will gain access, significantly expanding its application in national security and cyber intelligence.

Fable 5: Waiting and a New Control Architecture

The fate of Fable 5 remains uncertain. Unlike Mythos, this model was previously open to all subscribers and considered the most powerful public AI tool. However, its launch has now been postponed, with no specific timeline provided. The situation is complicated by a new executive order dated June 2, which introduces a voluntary federal review channel for advanced AI models. Developers are required to submit their solutions for cybersecurity checks 30 days before launch.

Notably, OpenAI has followed a similar path, restricting access to the most powerful version of GPT-5.6, named Sol, to just 20 state-approved partners. The weaker versions, Terra and Luna, have become publicly available.

Global Context and Geopolitics

The root of the blocks lies in fears that the technology could reach China. An incident involving South Korean operator SK Telecom—first added to the Glasswing list and then restricted—only confirms these risks. Leaders of major cybersecurity companies, including Nvidia, Adobe, and Zoom, have urged authorities to abandon overly restrictive measures, but Washington continues to tighten control.

My analysis: Lifting restrictions on Mythos 5 is just the first step toward shaping a new AI control architecture. In the coming days, we will see whether Fable 5 receives similar approval. However, it is clear that access to the most powerful models will now be strictly regulated by the state, and countries in Europe and other regions will become heavily dependent on Washington's decisions. The AI market is entering an era of "permitted" access, and this will change the rules of the game for all participants.