Sources

These are the families of public sources the posts are built from, with examples of each. It is not the complete list: the internal curation is broader and changes with what comes up. Some sources are read every day and others, like international bodies and research centers, in a weekly sweep.

Aggregators and the tech press are useful for learning about a story, but the citation goes to the original: the document, the regulation, the press release or the paper. When an outlet republishes a wire story, we cite the wire agency. Figures a company publishes about itself are presented as the company's figures.

How close to the facts each source is

Every post lists its key sources at the bottom, each with one of these four levels.

Level 1, primary or official. The document, the regulation, the ruling or the body that decides. A company is a primary source for what it announces, not for how well it works.
Level 2, independent specialist. Research, academic centers, civil society and evaluators who don't sell what they analyze.
Level 3, press. Outlets that report, cross-check and sometimes get there first. Used with attribution.
Level 4, interested party. Whoever is talking about their own product or business: their figures are flagged as such.

Frontier AI labs

Announcements and research from the people who build the models. Includes the Chinese labs, which set much of the pace of releases and show up often in the posts. These are company communications: they confirm what was released, not whether it is the best.

Universities and research centers in Latin America

What is being researched and debated about AI from the region.

Regional governance: ministries and regulators

National strategies, data protection authorities and regulators across the region. We follow several; these are some of them. When a post says something about a law or an official decision, the link goes here and not to the news story that covered it.

Multilateral bodies and standards

Global governance, regional cooperation and technical standards.

Civil society and tech policy

Organizations that research and monitor how AI is regulated and used. These work mostly on the global, European and US debate; the region's organizations are listed separately, further down.

Latin American civil society

Organizations in the region that research surveillance, data protection, platforms and AI regulation. They work on the region's cases and laws, and often get to what is happening in each country before the press does.

AI safety and evaluation

Those who measure the capabilities, risks and behavior of models.

Analysis and industry

Voices that explain the business, the trends and the use cases.

Specialized tech journalism

Outlets that cover the industry daily. They are in English, and much of the day's news appears there first. They are useful for keeping up; when there is an original announcement or document, the citation goes to that original.

The region's press

Latin American outlets with good coverage of technology and digital policy, along with the newsrooms that do investigation and fact-checking on platforms, elections and disinformation in the region. When an outlet republishes a wire story, we cite the wire agency.

Infrastructure, chips and energy

The physical side of AI: data centers, semiconductors and how much electricity all of this uses. Nvidia and TSMC are interested parties, because they sell what they describe, so we use them to know what was announced and what is being manufactured, not to measure the size of the phenomenon. That is what the IEA and SemiAnalysis are for, since they don't sell chips.

Curated research

Papers already filtered for relevance, easier to follow than raw arXiv.

List last reviewed on September 28, 2026.

Know a source we should follow?

Suggest a blog, a news outlet, a research center or an account you think adds something. We review every suggestion: if it fits what we do, we add it to the list; if not, we pass; and when it's a judgment call, it goes to human review.

Leaving your email is optional: we only use it if we need more context about the source or to let you know if we add it. It doesn't subscribe you to the newsletter, and we don't share it.