Lifting Export Restrictions on Claude Mythos 5: A New Era of AI Safety or Playing with Fire?
The US administration has officially lifted the two-week ban on the export of Anthropic's advanced artificial intelligence model, Claude Mythos 5. This decision opens access to the technology giant for over 100 American organizations, including major corporations and government agencies listed in the special Appendix A registry.
Commerce Secretary Howard Lutnick, in a letter to Anthropic's technical director Tom Brown, confirmed that a separate license is no longer required to transfer the Mythos 5 model to partners on the approved list. "I have concluded that current safeguards allow certain verified partners to access the Claude Mythos 5 model," Lutnick stated, emphasizing that security control mechanisms are deemed sufficient.
History of the Standoff: From Block to Unblock
Recall that restrictions on Mythos 5 and its more powerful version, Fable 5, were imposed 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. For two weeks, Anthropic's top managers held intensive negotiations in Washington, which ultimately led to the unblocking.
Notably, until this point, Mythos was used exclusively within the secret Glasswing project—a vulnerability search program involving about 150 organizations from over 15 countries. Previously, the Mythos model identified critical errors in classified government systems within hours during testing, further confirming its enormous potential.
Fable 5 Remains in the Shadows: New Rules of the Game
However, while Mythos 5 has received the "green light," the fate of Fable 5—a version previously available to all subscribers and considered the most powerful public AI tool—remains uncertain. According to available information, the release process has moved forward, but no specific timelines have been announced.
The situation is unfolding against the backdrop of a new regulatory framework. 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. This step, along with tightened export controls on AI chips to China, signals the formation of a fundamentally new approach to controlling technology dissemination.
Notably, OpenAI has followed the same path, restricting access to the most powerful version of GPT-5.6, named Sol, to about 20 state-approved partners. Weaker versions, Terra and Luna, have become available to the general public.
Geopolitical Context and Market Reaction
Initially, the blocks were driven by fears that competitors, particularly China, could acquire the technologies. The focus fell on South Korean operator SK Telecom, which was added to the Glasswing list in early June and then had its access restricted. SK Telecom itself denied any connection to China.
Many leaders of major cybersecurity companies, including former Facebook security chief Alex Stamos, as well as executives from Nvidia, Adobe, and Zoom, urged authorities to abandon restrictive measures. European countries and other regions have already noted their heavy dependence on Washington's decisions regarding access to new developments.
My analysis: The unblocking of Mythos 5 is not just a technical decision but a strategic move demonstrating that the US is ready to use advanced AI models as a tool of "smart power." However, the fate of Fable 5 will be the litmus test: can the industry maintain a balance between innovation and security, or are we entering an era where access to the most powerful algorithms will be determined solely by political loyalty? For the crypto and blockchain sector, where AI is increasingly used for smart contract auditing and vulnerability analysis, these events have direct implications—access to the best tools may become a privilege, not a right.