Sistema Alerta Niñez — Chile's Machine-Learning Score for Children at Risk of Rights Violations

Chile's Sistema Alerta Niñez (SAN) is a machine-learning risk-scoring tool built by the Subsecretaria de Evaluacion Social within the Ministry of Social Development and Family. Since 2019 it has combined administrative and self-reported data -- covering family, educational, social and territorial variables -- to estimate which children and adolescents face the highest risk of having their rights violated.

To let Local Childhood Offices (Oficinas Locales de la Niñez) prioritise early intervention for children and adolescents at highest risk of rights violations, such as abuse, school dropout, substance use or exploitation.

The system has run continuously as a pilot in 12 communes spanning multiple regions -- from Iquique in the north to Aysén in Patagonia, including Santiago, Concepción and La Florida -- combining administrative and self-reported data to generate a risk score for each child.

By October 2019 the system had helped direct services to 2,262 children and adolescents, against an estimated annual intervention capacity of roughly 3,440 in the participating offices. Despite an original ambition to expand nationwide to all 345 Chilean communes, SAN remained confined to the same 12 pilot communes as of 2024-2026.

A 2022 investigation by the digital-rights organisation Derechos Digitales found that the procurement process that created SAN set no ethical, transparency or data-justice criteria in advance, and that no public documentation exists of the model's design, variables or accuracy. Civil-society and academic reviewers have raised concerns about stigmatisation risk and the absence of any independent evaluation of whether the tool actually reduces harm to flagged children, rather than simply reallocating limited casework capacity.