The block on Anthropic’s Fable 5 and Mythos 5 models changed in nature today: it is no longer just a matter of export controls, but a national security issue with a political way out in sight. The US National Cyber Director, Sean Cairncross —part of the National Security Council, not of the commerce office that imposed the restriction—, formally joined the negotiations, and the newspaper The Globe and Mail reported this morning that there is a “deal in the works” with a possible resolution before June 30 (status of the block · report on the deal). As a reminder: for eleven days these advanced models have been banned for part of the world, including Latin America, in an episode we have been following closely.
That an official from the National Security Council —and not from the Bureau of Industry and Security (BIS), which applied the ban— is sitting at the table raises the case to the highest political level and suggests that Washington is setting the standard for who can access frontier AI models, systems capable of tasks so sophisticated that they are considered sensitive for defense. For the region there is an uncomfortable detail: if restoring access requires citizenship verification or biometric data —something hinted at by Anthropic’s June 19 privacy policy update—, Latin American users could be left out of a solution designed for US residents. Yesterday, in addition, a commercial change took effect without notice: Fable 5 stopped being included in the subscription and began to be charged through additional credits. Prediction markets give a 57% probability that the block will be resolved before July 1.
Also today
- Gemini 3.5 Pro enters general availability this week — Google’s new model, with a context window of two million tokens, opens its general availability between June 23 and 30. It arrives while Fable 5 remains blocked and, with Android on more than 80% of phones in the region, it is the frontier model that reaches the most Latin American users by default.
- Colombia moves forward with the certification of its runoff — The June 21 result —De la Espriella 49.66% against Cepeda 48.70%— still awaits formal ratification, with the official count scheduled for midweek. The state documented digital impersonations on election day itself, but its anti-deepfake law only begins to impose sanctions in July.
- GPT-5.6 seems imminent, without official confirmation — Prediction markets raise to 83% the probability of an OpenAI announcement before June 30, according to secondary sources. It is best to wait for the official announcement before taking it as fact.
In the region
In 48 hours, UNESCO’s Third Ministerial Summit on the Ethics of AI in Latin America and the Caribbean opens in Santo Domingo, organized together with the Dominican OGTIC and the development bank CAF (Foro IA LAC). It is the first regional ministerial forum held with the Fable 5 block in place and with the Colombian result on the table, and the first multilateral setting with a Latin American voice since the Évian G7 closed without a common AI framework. The ministers arrive with a clear map of two gaps: the access gap —decided unilaterally in Washington— and the electoral regulation gap, where Colombia showed that manipulation can be detected in real time but not sanctioned in time. The practical question is whether the summit dares to announce a concrete, enforceable instrument before Brazil’s elections in October.
Threads we’re following
The Colombian outcome is one of the threads we have been carrying for weeks. We had been reporting that its runoff was the region’s most closely watched election in the face of AI disinformation, decided by less than one point and with no way of measuring how much weight the six deepfakes documented during the campaign carried. Today’s chapter adds a twist: President-elect Abelardo de la Espriella outlined his technology priorities for 2026-2030 —AI for the state, blockchain and an alliance with SpaceX for rural connectivity (El Tiempo). The paradox is plain to see: the same conglomerate linked to the Fable 5 ban would appear as an infrastructure provider for Colombia’s next government.
If the way out of the Fable 5 block ends up requiring citizenship verification, what is left for Latin America: wait for the version Washington approves, or build its own access to frontier models?
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.