What changed today: Colombia now has a result, and it is the one no one predicted. Abelardo de la Espriella won the presidential runoff with 49.66% against 48.70% for Iván Cepeda —a margin of less than one point, when polls gave De la Espriella a seven-point lead— with a turnout of 63.6% (Al Jazeera). On election day itself, the Technology Ministry (MinTIC) published findings on coordinated digital impersonation campaigns that were still active while the polls were open (MinTIC). We had been following this campaign because it exposed a state capable of detecting disinformation in real time, but not of sanctioning it: Law 2502, which provides the mechanism to punish deepfakes —fake videos or audio fabricated with artificial intelligence—, does not take effect until July 2026.
Such a tight result makes the experiment’s underlying question impossible to answer. With six documented fabrications in the campaign and a final margin of 0.96 points, there is still no methodology to estimate how many votes each piece of disinformation moved. The central unknown closes without an answer just as voting ends, and that is precisely what makes the Colombian case so instructive: it is not a model of regulatory success, but the most detailed map the region has of the gap between what a state can see and what it can prevent.
That distinction matters because Colombia does not vote alone in this cycle. Brazil goes to the polls in October 2026, and Chile and Argentina have their own calendars ahead. The legacy Colombia hands over is neither reassuring nor catastrophic: it is concrete. It showed that it is possible to put together electoral monitoring deployed on several fronts and that such monitoring works for observing and warning. And it showed, just as clearly, that detecting without being able to sanction does not stop the erosion of democratic trust; it documents it precisely.
In the region
In three days, on June 25 and 26, the region’s ministers in the field meet in Santo Domingo, and they arrive with the most concrete material available so far for discussing electoral frameworks: a presidential election decided by less than one point in which the state detected disinformation operations in real time but could not apply a single sanction. The discussion stops being abstract. The practical question Colombia puts on the table is whether the region can close the timing gap —the distance between the capacity to detect and the capacity to sanction— before the next major vote, and not after.
Also today
- OpenAI’s GPT-5.6 is rumored for this week — Leaks on specialized blogs point to Mini and Pro versions with better reasoning and a voice mode during the week of June 22. OpenAI has made no official announcement, so it is best treated as a rumor until the company confirms it.
Threads we’re following
The block on Anthropic’s Fable 5 model, which we have been following since the middle of the month, reaches its tenth day today and brings two new developments. The first is technical and reframes everything: US Senator Mark Warner cited the director of the National Security Agency (NSA), General Joshua Rudd, as saying that the restricted Mythos model accessed almost all of the United States’ classified systems “within hours” during an authorized security exercise (Anthropic). It is the first high-level technical justification for the urgency of the ban —though it is worth clarifying that it is Warner’s account of what Rudd said, not a direct statement from the agency. The paradox for Latin America remains intact: the blocked model is not Mythos, which is restricted-use, but Fable 5, a general-purpose model, and users in the region have gone ten days without access to the latter because of the capabilities of a model they were never able to use. The second development is commercial: today the change takes effect that moves Fable 5 from being included in subscription plans to being charged through credits, with no public communication from Anthropic about what happens to affected subscribers.
If the ministers meeting in Santo Domingo wanted to leave with a single concrete, enforceable instrument before Brazil’s elections in October, would it be an electoral framework with the capacity to sanction in real time, a regional consultation protocol on decisions made in Washington, or something we have not yet managed to name?
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.