El Salvador Embarks on a Perilous A.I. Experiment

Leer en español

In the 1940s, researchers from the United States and Guatemala infected hundreds of Guatemalans with syphilis and gonorrhea in experiments to test vaccines. Decades later, the Pan American Health Organization apologized, calling the studies “reprehensible” and “deeply troubling.” The work had been funded by its predecessor agency.

The differences with today are obvious. No one is conducting medical experiments on human bodies. Yet a similar logic is emerging in a different realm: data.

In both cases, vulnerable populations become the raw material for projects designed elsewhere, by actors with far greater power and resources.

El Salvador shows how artificial intelligence and data infrastructure are expanding in countries with weak checks on political and corporate power and a heavy dependence on foreign technology firms. It also illustrates how A.I. can become intertwined with authoritarian governance.

Five years ago, President Nayib Bukele made El Salvador the first country to hastily adopt Bitcoin as legal tender, years before passing legislation aimed at strengthening personal data protections. The project eventually flopped: Bukele revoked Bitcoin’s legal tender status, under pressure from the International Monetary Fund and low public adoption.

This year, El Salvador has swung open the doors of its public institutions to large technology companies, often on generous terms that include substantial tax exemptions.

[rel2]

Major technology companies appear to be on board, including Google and Nvidia. Their involvement coincides with reports that the Trump administration is seeking access to Africans’ personal data in exchange for health cooperation under the America First Global Health Strategy, an agreement signed by 30 countries, including El Salvador, Honduras and Guatemala.

One example is Doctor SV, an application built on Google's Gemini A.I. system and presented jointly by Bukele and Google. The government says the platform will improve access to health care, strengthen medical services and handle as many as 30,000 consultations a day. Other A.I. initiatives announced by the government touch public education and criminal prosecution.

These projects, on the heels of El Salvador’s February 2025 A.I. Law, have garnered support from the World Bank, IMF, and Trump administration.

The Doctor SV initiative is part of a broader agreement between El Salvador and Google reportedly worth about $500 million over seven years and financed in part through international loans. The deal has been classified as a state secret, limiting public scrutiny.

What is known is that it involves the deep integration of government services with private technological infrastructure, including access to sensitive health and administrative information.

[newsletter]

Doctor SV sits at the center of that partnership. It links hospitals, laboratories, pharmacies, and call centers through Google's suite of services. Bukele has described it as a system “unlike anything in the world.” The Pan American Health Organization has also expressed support for the project, drawing concern from some observers given the agency's historical connection to the Guatemala experiments.

The result is a growing concentration of sensitive personal information within infrastructure controlled by a private corporation. Officials argue that patients consent to the use of their data. In practice, however, consent is difficult to assess when transparency is limited and decision-making is highly centralized. Under those conditions, legal safeguards often matter less than political discretion.

The larger question is what kind of model this represents. One vision of the A.I. revolution is straightforward: invite major technology companies into the machinery of the state and grant them access to vast quantities of public data, with little public debate over how that information is used or whether it ultimately contributes to commercial A.I. systems.

A separate initiative with Nvidia reportedly seeks to create a synthetic population of one million “digital citizens” designed to mirror El Salvador's demographics and help train A.I. systems. At the same time, the government has reduced public-sector employment in areas such as health care and education amid austerity pressures linked to international financial institutions led by the IMF.

[rel1]

The scholars Nick Couldry and Ulises A. Mejias describe this pattern as “data colonialism”: the transformation of human life into data that can be extracted and monetized, much as land and natural resources were during earlier eras of colonial expansion.

For Valeria Di Croce, a professor at the National University of La Plata who studies the relationship between Javier Milei and artificial intelligence companies, a similar dynamic is unfolding in Argentina.

“The goal is not to develop our own technology companies, but to integrate ourselves into the global supply and infrastructure chain — the new Potosí,” she told me, referring to the historic Bolivian mining city whose wealth fueled the Spanish Empire.

Argentina, she noted, offers resources particularly attractive to the A.I. industry: vast tracts of land, Patagonia’s cool climate, lithium-related minerals and abundant freshwater.

That contrasts with El Salvador, whose limited territory and growing water stress leave it with fewer opportunities to serve as infrastructure for the A.I. economy. Instead, its value lies increasingly in its institutions, its data and its willingness to provide a permissive environment for technological experimentation on people.

Viewed this way, the issue is not simply technological innovation. It is about the relationship between A.I. companies and governments with few democratic constraints. Supporters point to efficiency, speed, and modernization.

The more consequential questions concern who controls the data, who profits from it, and who is excluded from the decisions surrounding its use. In a country controlled by a single leader, citizens cease to be participants and quickly become test subjects.

Ricardo Valencia is an associate professor of communications at California State University, Fullerton. Find him on X: @ricardovalp