The US government has lifted the mandatory licensing requirement for exporting the latest artificial intelligence models, Mythos and Fable, developed by Anthropic. Previously imposed sanctions effectively blocked access to cutting-edge technologies outside US jurisdiction, leading to the forced shutdown of services for a wide audience. This is reported by RBK-Ukraine citing a TechCrunch article.
From Sanctions to Retreat
Strict restrictions were introduced by the Trump Administration on June 12, when Anthropic products were officially added to the list of dual-use technologies. In practice, controlling foreign access to cloud models proved impossible: the company was forced to completely cut off access to its advanced AI for international users.
Cybersecurity experts immediately labeled this move as political retribution by the government for criticism of the authorities by the startup's top managers. However, the White House was forced to retreat urgently due to geopolitical factors and market reality.
Asian Breakthrough and Loss of Competitiveness
Technological giants from Asia began releasing their own neural networks — specifically, Fugu and Tulongfeng, which came close to the capabilities of the American Mythos. The strict ban by the US government simply deprived American companies of global competitiveness, handing the initiative to foreign developers.
Compromise and New Rules of the Game
Commerce Secretary Howard Lutnick stated that Anthropic agreed to proactively identify security risks, closely cooperate with the government on standards for future releases, and immediately report any harmful user activity. This became the basis for a compromise deal that allowed the export ban to be lifted.
It is worth noting that even before the introduction of sanctions, Anthropic voluntarily implemented strict security protocols. The Fable version was initially released with additional internal restrictions to minimize the risks of searching for and exploiting software vulnerabilities.
Chaos in White House Policy
Despite the lifting of the export ban for Anthropic, the White House's overall policy towards AI remains chaotic and unpredictable. In recent weeks, the government has practiced selective releases: the most powerful versions of Mythos and the newest secret models from competitors at OpenAI are currently allowed to be supplied only to a limited circle of organizations personally approved by Donald Trump's team, rather than being opened to the general public.
The June presidential decree introducing mandatory preliminary review by government agencies of all AI models before their release sparked a wave of harsh criticism in the industry. Even leading analysts, in particular Dean W. Ball, who recently took office as Director of Technology Policy at OpenAI, note that such steps by the authorities deprive developers of clear guidelines and slow down the development of the secure infrastructure necessary for the future era of autonomous AI agents.
Thus, the lifting of the export ban on Anthropic models was less a victory for technological freedom and more a forced maneuver in the context of global competition and internal pressure from the tech community.
