Anthropic breaks through the blockade: Mythos 5 admitted to 100+ organizations in the US, Fable 5 at the starting line
The American regulatory deadlock regarding Anthropic's advanced AI models has finally been resolved. The U.S. Department of Commerce officially lifted the export ban on the Claude Mythos 5 model on Friday, opening access to it for over 100 American organizations. This list includes both the largest corporations and government agencies. The decision was the result of a two-week standoff between the administration and the developer, putting an end to the dispute over the safety and dissemination of next-generation technology.
Mythos 5: From Secret Project to Broad Access
Commerce Secretary Howard Lutnick confirmed that a special license is no longer required to transfer Mythos 5 to partners under "Appendix A." Previously, the model was only available within the Glasswing program — an initiative to search for vulnerabilities involving about 150 organizations from 15 countries. It is worth recalling that Mythos 5, within hours of testing, identified critical errors in classified government systems, which sparked the initial concerns.
Notably, the blockade was triggered not only by national security concerns but also by pressure from Anthropic's key investor, Amazon. It was the giant's fears regarding the potential vulnerability of the models that led to the temporary freeze. Now that the restrictions are lifted, Mythos 5 returns to its partners.
Fable 5: Waiting in the Shadow of New Regulation
The fate of Fable 5 — an even more powerful public model — remains in question. Unlike Mythos, access to it was previously open to all subscribers, and it was considered the strongest publicly available AI tool. However, its broad release is now postponed. The administration has not yet announced a timeline, but the process has clearly moved from a standstill.
It is important to note that the situation is unfolding against the backdrop of a new control system being formed. On June 2, an executive order was signed creating a voluntary channel for federal review of advanced AI models. Developers can submit their solutions for a cybersecurity check 30 days before launch. This effectively introduces a precedent of pre-approval for the most powerful algorithms.
In parallel, OpenAI followed a similar path on Friday, restricting access to its strongest version, GPT-5.6, named Sol, to approximately 20 state-approved partners. The weaker versions, Terra and Luna, became publicly available. The trend is clear: Washington is strictly controlling the dissemination of the most advanced models, fearing they might fall into the hands of China. The incident with the South Korean operator SK Telecom, which was temporarily added to the Glasswing program and then had its access restricted, only confirms these concerns.
Analytical Conclusion: The AI market is entering a new era where access to advanced models will be determined not only by technological capabilities but also by geopolitical loyalty. For the crypto industry, which actively uses AI for DeFi protocols and analytics, this means a potential slowdown in innovation and increased dependence on Washington's decisions. Whether Fable 5 gets the green light will be a litmus test for the entire sector.