Anthropic has just opened an office in Seoul —its third location in Asia-Pacific, after Tokyo and Bengaluru— together with Korean tech giants such as NAVER, Samsung SDS and LG CNS. And it does so amid a paradox that is hard to overlook: the partners signing those agreements cannot use Fable 5, the company’s most capable model, because it remains blocked worldwide by a US export control (TechTimes). As context: a week ago the US government “switched off” the Claude Fable 5 and Mythos 5 models across the planet after a way to get around their safety barriers was detected, and the ban remains in force with no date for resolution.
The block reaches its eighth day today and, for the first time, gains a concrete deadline that does not depend on Washington. Until June 22, access to Fable 5 came included at no cost in Anthropic’s paid plans (Pro, Max, Team and Enterprise); starting on the 23rd it will be charged as additional usage credits (TechTimes). In other words, the commercial calendar moves forward even though the political one is frozen. The first sign of optimism with a name and title attached came, precisely, from Seoul: KiYoung Choi, head of the new office, said he was “very confident” that the models will return “in the coming days,” though without giving a date. The underlying bet is clear: Anthropic signs alliances with the future architecture in mind, not today’s access.
For Latin America —which also had no seat in any of these decisions— the reading is the same as all week, now with a commercial nuance. The region’s access to the frontier of artificial intelligence is not defined by which model performs best on tests, but by a map of interests between Silicon Valley and Washington; and while that map sorts itself out, even the terms of the subscription a user pays for in Bogotá or Santiago can change overnight without the ban having been lifted.
Also today
- Colombia reaches the eve of the runoff with monitoring on five fronts — the Technology Ministry (MinTIC) and the ColCERT response team deployed the region’s most sophisticated electoral surveillance for tomorrow’s vote, June 21. They can detect and warn, but not sanction: the law that punishes deepfakes does not take effect until July.
- Newtral documents active hoaxes hours before the close of the campaign — fake “evidence” of fraud and videos taken out of context circulating this very morning in Colombia, according to the Spanish fact-checker.
In the region
Colombia votes tomorrow with the region’s most documented electoral AI experiment coming to a close without the capacity to sanction: six deepfakes accumulated during the campaign, a ministry monitoring hour by hour and zero possible fines until July. The precedent it leaves —which Brazil, voting in October 2026, as well as Chile and Argentina, will watch closely— is not a model of a successful response, but a map of what remains to be built before the next cycle. At the same time, Brazil is starting to show the other side: the data protection authority (ANPD) already has three companies —Metatext, Synapse and IA Greenworld— in an active testing phase within its AI regulatory “sandbox,” a controlled environment for trying out rules with a focus on algorithmic transparency, in place until December 2026 (IAPP). It is the first real regulatory move in Latin America with specific companies, and the kind of quiet precedent that other regulators in the region are going to study.
Threads we’re following
We had been following the first government “shutdown” of frontier AI models: the ban with which the United States cut off global access to Claude Fable 5 and Mythos 5, a story that went from a technical dispute to a personal clash and from there to the geopolitical board. Today’s chapter adds two new things: a sign of optimism from inside Anthropic —the first with a name and title attached— and a commercial clock of its own, the June 22 one, which runs independently of the political negotiation. The Korean expansion shows that the company keeps betting on the long term even with its best model blocked.
If access to the world’s best AI can be switched on and off according to a dispute between Silicon Valley and Washington, and the regulatory precedents that matter most to us are being written this week in Bogotá and São Paulo, will Latin America arrive at the next ministerial summit to describe what happened to it, or to decide what it is going to do?
About this entry. It is generated automatically from public sources, without human review before publication. It may contain errors of interpretation or summary; please check each story against its original source (the links lead there) before citing it or making decisions based on it.
Doble Click is written with Anthropic models.