Anthropic breaks through the blockade: Mythos 5 gets the "green light" for the top 100 US organizations, Fable 5 on the way
On Friday, the U.S. Department of Commerce officially lifted export restrictions on Anthropic's Claude Mythos 5 model. This decision unlocked access to the advanced AI tool for over 100 American organizations, including major corporations and government agencies.
This move ended a two-week standoff between the Trump administration and Anthropic. Previously, exports of Mythos and its "younger sibling" Fable 5 were suspended due to concerns raised by one of the key investors — Amazon. Company researchers warned that Fable 5 could be hacked for potentially dangerous purposes, which triggered the temporary restrictions.
Mythos 5: From Glasswing to widespread adoption
Until today, Mythos 5 was used exclusively within the Glasswing program — a vulnerability discovery initiative involving about 150 organizations from 15 countries. Notably, this model identified critical errors in classified systems within hours during government testing. Now, according to a letter from Commerce Secretary Howard Lutnick to Anthropic's CTO Tom Brown, a license is no longer required to transfer Mythos 5 to organizations listed in Appendix A.
"The existing safeguards allow certain verified partners to access the Claude Mythos 5 model," Lutnick noted, confirming that the control mechanisms are deemed sufficient.
Fable 5: Waiting in a new reality
The fate of Fable 5 remains less certain. Unlike Mythos, access to this model was previously open to all subscribers, and for a time it was considered the most powerful public AI tool. However, its release is now delayed, with no specific timeline announced. The situation is compounded by an executive order signed on June 2, which opened a voluntary channel for federal review of advanced AI models. Developers can now submit their solutions for cybersecurity checks 30 days before launch.
Notably, OpenAI followed a similar path on the same Friday, restricting access to the most powerful version of GPT-5.6, named Sol, to about 20 state-approved partners. Weaker versions, Terra and Luna, were made available to the general public. This trend of stratifying AI access is becoming dominant.
The root of the blockages was concern that the technology could fall into the hands of competitors, primarily China. Attention was 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. Many 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.
My analysis: Lifting the blockade on Mythos 5 is not just a victory for Anthropic, but a signal of the formation of a new architecture for global AI control. We are witnessing the birth of a "two-tier" system: elite models for verified partners and "weakened" versions for the mass market. Fable 5 will likely receive similar approval, but its public status may be reconsidered in favor of stricter regulation. Europe and other regions have already realized their dependence on Washington's decisions, which will only heighten geopolitical tensions in this area.