French startup Mistral AI is in talks to raise about 3 billion euros at a valuation close to 20 billion, as revealed by Bloomberg. The figure nearly doubles what it was worth just nine months ago, and it comes at a very particular moment: the round is being negotiated just as Anthropic’s two most advanced models remain switched off across the planet by order of the US government. Mistral, which presents itself as the “sovereign” European option against the US duopoly, thus finds a sales argument handed to it on a silver platter.
The idea of technological “sovereignty” stopped being abstract this week. Sovereignty, here, means something very concrete: that no outsider can switch off the infrastructure you depend on. Europe has spent days insisting that it wants to make sure “no one has a kill switch” over its digital tools, and Mistral is the corporate face of that ambition. The detail that makes all this relevant for Latin America is that Mistral already operates commercially in Colombia, Chile, Mexico, Brazil and Argentina. In other words: the region has at hand, today, an alternative that does not depend on Washington, at the very moment it found out what it costs to depend on it.
It is worth not idealizing it: a higher valuation does not by itself mean better models or real independence, and raising money at that scale also ties Mistral to its own investors and geographies. But the move sketches a more plural map, where frontier AI stops being an exclusively US affair and begins to have European accents —and, by way of commercial availability, Latin American ones too.
Also today
- The US Department of Health uses ChatGPT to track up to $200 billion in Medicaid fraud — The AERO program audits five years of records in all 50 states and becomes the first federal-scale deployment of generative AI on social programs in that country. A signal of where the use of these tools in public administration could be heading.
- Anthropic says 80% of its code is already written by Claude and calls for a coordinated global pause — In the report “When AI Builds Itself,” the company argues that AI already takes part in the development of AI itself and proposes slowing down in a coordinated way. The proposal coexists with the underlying tension of the whole industry: competing and, at the same time, calling for caution.
In the region
Today, June 15, voting abroad begins for Colombia’s June 21 presidential runoff: about 1.4 million Colombians in 67 countries will be able to vote over seven days, according to the National Registry. Registrar Hernán Penagos again warned about disinformation as a direct threat to democracy, while Law 2502 against deepfakes will not effectively take effect until July: the election arrives before the rule designed to protect it. At the same time, Ecuador joined the regional governance map by publishing, together with UNESCO, its readiness assessment for ethical AI —29 recommendations—, and becomes the fifth country in Latin America to complete this analysis, along with Brazil, Chile, Mexico and Guatemala.
Launches
- iOS 27 rebuilds Siri on a Google Gemini model — Apple announced at its developer conference that the new Siri will run on a customized Gemini model. There is a beta version for developers, and the public launch is expected in the second half of 2026. It is worth watching the quality of its answers in Spanish and how well it carries out actions within apps.
Threads we’re following
The backdrop to all this is the story we have been following since June 12: that day, an order from the US Department of Commerce forced Anthropic to switch off its Claude Fable 5 and Mythos 5 models for the whole world, in the first case of a government using an export control as an off switch on a frontier AI model. Three days later they still have no return date; presidential AI adviser David Sacks backed the directive, while Anthropic insists that the “jailbreak” that prompted the measure is narrow and that the standard, if generalized, would halt all of the industry’s deployments. That void —that of a region that overnight lost access to tools it used daily, without having been consulted— is exactly the gap that an alternative like Mistral aims to fill.
If the lesson of the week is that AI infrastructure can be switched off from afar, is it enough to change providers, or does real sovereignty require building the region’s own capabilities?
Correction (September 30, 2026). The original headline said “Mistral negotiates 20 billion”; the correct headline is “Mistral negotiates a valuation of 20 billion”: that figure, in euros, is the valuation, and the round is about 3 billion euros, according to TechCrunch, citing Bloomberg.
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.