The AI sector moves quickly, and a lot has happened in a week. Last Tuesday, Anthropic released Fable 5, a restricted version of the Mythos model that the company said in April was too dangerous to release to the public. Due to the risks that it could be used for cyberattacks or to develop bioweapons, Fable comes with significant safeguards.
Despite those limits, the Trump administration on Friday placed a form of export control on the technology, blocking foreign citizens—including those inside the United States and even those working for Anthropic—from accessing it. To ensure compliance with such a sweeping order, Anthropic disabled access to all “Mythos-class” models for everyone. The shutoff, while likely temporary, highlights the ongoing debate over the pros and cons of restricting these AI models as if they were weapons—a topic my fellow columnist Paul Poast wrote about last Friday.
For the narrow question of regulating the most powerful frontier models, Latin America is largely a spectator. The region neither develops nor hosts the most powerful models. If anything, its interests run counter to Washington’s and would benefit from gaining access to more powerful technology that the U.S. is trying to fence off.