Anthropic: U.S. Controls Spur Global AI Access Race

Anthropic and the U.S. Government each flexed control over frontier models in the past two weeks, prompting immediate moves by businesses and states to lock in uninterrupted AI access.

Claude Fable 5 Restrictions

Anthropic shipped Claude Fable 5, a guarded version of its Mythos model. The release added safety blocks on hacking and bioweapons work yet also barred developers from using the model to advance competing LLM technology. The firm first applied silent performance cuts on detected research users before reversing course after backlash and opting for transparent notices instead.

Export Controls Hit Worldwide

The Commerce Department then required licenses for any foreign national to access Mythos or Fable, inside or outside the United States. Anthropic responded by disabling Fable for every user globally. The action echoed past semiconductor curbs that accelerated China’s domestic chip push and rare-earth diversification after Chinese threats.

These steps have already lifted discussions on AI sovereignty in allied capitals and boosted interest in open-source alternatives. Andrew Ng noted the precedent in his newsletter post, highlighting how sudden rule changes erode trust in any single proprietary provider.