Anthropic breaks through export blockade: Mythos 5 gets "green light" for 100+ US organizations, Fable 5 on the horizon
The Trump administration has lifted the two-week ban on the export of Anthropic's flagship model, Claude Mythos 5. This decision, officially formalized on Friday by the U.S. Department of Commerce, opens access to the advanced AI tool for over 100 American organizations, including major corporations and government entities listed in the special "Appendix A" roster.
Commerce Secretary Howard Lutnick confirmed in a letter to Anthropic's Chief Technology Officer Tom Brown that existing precautionary measures are deemed sufficient, and a separate license is no longer required to transfer Mythos 5 to trusted partners. This marks the end of a tense standoff between the White House and the company, which arose after Amazon — one of Anthropic's key investors — expressed serious concerns about the safety of both Mythos and its more powerful "counterpart," Fable 5.
Recall that access to Mythos was previously strictly limited within the framework of the secret Project Glasswing — a vulnerability discovery program involving about 150 organizations from 15 countries. Mythos demonstrated its incredible effectiveness in just a few hours, identifying critical errors in classified government systems during testing. Now, with the lifting of the block, its potential will be unlocked for a much wider range of users within the United States.
Fable 5: Waiting in the Shadow of New Regulation
However, the fate of the even more powerful model, Fable 5, remains in question. Unlike Mythos, which was previously available only to a select few, Fable 5 was originally positioned as the most powerful public AI tool for all subscribers. Its broad launch is now postponed indefinitely.
Concurrently, a new control system is taking shape in the U.S. On June 2, an executive order was signed, opening a voluntary channel for federal review of advanced AI models. Developers can now submit their solutions for a 30-day cybersecurity check before launch. This step, along with tightening restrictions on AI chip exports to China, signals a shift toward a more structured, yet more restrictive, regime.
OpenAI has already followed this trend, limiting access to the most powerful version of GPT-5.6, named Sol, to just 20 state-approved partners, while the weaker versions, Terra and Luna, have become publicly available. Washington's logic is clear: to prevent the leakage of the most advanced technologies to geopolitical adversaries, especially China. The incident involving South Korean operator SK Telecom, which was temporarily restricted from accessing Glasswing, only confirms this paranoia.
My view: The unblocking of Mythos 5 is a tactical victory for Anthropic and a signal to the market that dialogue with regulators is possible. But the story of Fable 5 clearly demonstrates: the era of unfettered distribution of frontier models is over. We are entering a period of "managed AI," where access to the most powerful algorithms will be a privilege, not a right, and will be tightly tied to Washington's political and economic interests. The only question is how effective this system will prove against truly determined players.