CHL-002

Sistema de Alerta Ninez (SAN) -- ML-Based Child Welfare Risk Prediction

Download PDF
Chile Latin America & Caribbean High income Pilot / Controlled Trial Phase Confirmed

Ministerio de Desarrollo Social y Familia (MDSF); Subsecretaria de Evaluacion Social (system owner); Subsecretaria de la Ninez (operational delivery via OLN offices)

At a Glance

What it does Prediction (including forecasting) — Vulnerability, needs and risk assessment, including predictive analytics
Who runs it Ministerio de Desarrollo Social y Familia (MDSF); Subsecretaria de Evaluacion Social (system owner); Subsecretaria de la Ninez (operational delivery via OLN offices)
Programme Oficinas Locales de la Ninez (OLN) -- Child Protection Services under the Subsecretaria de la Ninez
Confidence Confirmed
Deployment Status Pilot / Controlled Trial Phase
Key Risks Governance and institutional oversight risks
Key Outcomes 2,262 children received intervention as of October 2019 across 12 pilot comunas.
Source Quality 3 sources — Government website / press release, News article / media, Report (multilateral / development partner)

Chile's Ministerio de Desarrollo Social y Familia (MDSF) developed and deployed the Sistema de Alerta Ninez (SAN), publicly known as Alerta Infancia, as part of the Acuerdo Nacional por la Infancia announced in May 2018. The system uses machine learning techniques applied to cross-referenced administrative data from multiple public institutions to estimate the risk that individual children and adolescents will suffer violations of their rights. It produces a prioritised list of at-risk children for the Oficinas Locales de la Ninez (OLN), local children's offices operated by the Subsecretaria de la Ninez, enabling earlier and more targeted intervention.

The system cross-references data that multiple public institutions collect about vulnerable minors, spanning health status, education, family context, psychosocial information, socioeconomic history, and environmental and territorial factors. Data sources include administrative databases from multiple government agencies as well as self-reported data. Additionally, field workers (sectoralistas) serving children and families can input alerts when they observe risk factors in the community, providing a human information layer alongside the algorithmic outputs.

The technical approach is described in official documentation as using advanced data analytics techniques (tecnicas de analitica avanzada) and machine learning to generate vulnerability indicators and a risk estimation for each child. OLN professionals -- coordinators, case managers (gestores), and therapists -- use the prioritised information to deliver a more comprehensive and timely response, aiming to intervene before rights violations occur rather than reacting afterwards.

The system was implemented as a pilot in 12 comunas (municipalities) by October 2019: Iquique, La Serena, San Felipe, Requinoa, Cauquenes, Concepcion, Nueva Imperial, Aysen, Santiago, La Florida, Colina, and Quillon. As of that date, 2,262 children had received the intervention, with an estimated annual capacity of approximately 3,440 interventions across the participating OLNs. The plan projected rollout to all 345 comunas of Chile within five years.

The system was a finalist in the 2019 Concurso Funciona, a government innovation competition administered by the Servicio Civil, indicating institutional recognition as a public sector innovation.

The system has attracted significant criticism. Francis Valverde, spokesperson for the Bloque por la Infancia and executive director of the Asociacion Chilena Pro Naciones Unidas (ACHNU), called it 'highly dangerous because what it is doing is marking children, stigmatizing them in practice,' questioning the criteria for determining which children are at social risk and whether adequate resources would follow the alerts. Pablo Viollier, a public policy analyst at ONG Derechos Digitales, stated that the data being collected is 'doubly sensitive' because it concerns minors who receive special protection and includes medical history, socioeconomic data, and psychosocial information which is 'very stigmatizing.' Patricia Pena, an academic at the Instituto de Comunicacion e Imagen at the Universidad de Chile, questioned how well-trained an algorithm can be to make automatic decisions about a person's life, stating that 'an algorithm does not know the details of what occurs in a family environment; the human component is necessary.'

A 2021 research report by Matias Valderrama for Derechos Digitales, supported by Canada's International Development Research Centre (IDRC), raised additional concerns including: strong technological determinism in the development process (system built first, institutional framework defined later), procurement oriented toward a single vendor with no ex-ante criteria for ethics or data justice, a socioeconomic gradient with overrepresentation of children from lower-income households, and lack of public documentation about model design and performance. The report was based on transparency law requests and secondary documentation, as MDSF explicitly refused to allow key actors to be interviewed.

No specific algorithm type, model architecture, accuracy metrics, or technology vendor has been publicly disclosed.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Case management primaryAssessment of needs/conditions + enrolment
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name Oficinas Locales de la Ninez (OLN) -- Child Protection Services under the Subsecretaria de la Ninez
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Other
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Programme Description ML-based predictive risk scoring system that cross-references administrative data from multiple Chilean government agencies to identify children and adolescents at risk of rights violations, enabling prioritised intervention through local children's offices.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Classical ML
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Integration and Deployment
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Not documented
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Not documented
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary Not assessed
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary High
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HITL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Not documented
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Governance and institutional oversight risks
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Not assessed
Documented Risk Events MDSF refused transparency law requests for interviews with key actors. Procurement process was oriented toward a single vendor with no ex-ante criteria for ethics, transparency, or data justice. System proactively contacts families who have not sought government help, raising concerns about intrusion into private life. Marked socioeconomic gradient with overrepresentation of children from lower-income households.

Impact Dimensions

Equality, non-discrimination, fairness and inclusion
  • Human oversight protocol
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsSpecial categoryLinks data across multiple systemsCurrently available and usedCross-references data from multiple government agencies spanning health, education, family context, psychosocial, and socioeconomic domains. Also incorporates field-gathered information from territorial workers. Data concerns minors and includes sensitive categories (medical history, psychosocial status).

Servicio Civil (2019) 'Sistema de Alerta Ninez', Concurso Funciona 2019 entry. Available at: https://www.serviciocivil.cl/funciona/ (Accessed: 30 March 2026).

View source Government website / press release

Diario y Radio Universidad Chile (2019) 'Alerta Infancia: el software que expone los datos personales de ninos y ninas en riesgo social', Diario y Radio Universidad Chile, 29 January. Available at: https://radio.uchile.cl/2019/01/29/alerta-infancia-el-software-que-expone-los-datos-personales-de-ninos-y-ninas-en-riesgo-social/ (Accessed: 30 March 2026).

View source News article / media

Valderrama, M. (2021) 'Inteligencia Artificial e Inclusion en America Latina: Sistema de Alerta Ninez', Derechos Digitales. Available at: https://www.derechosdigitales.org/ (Accessed: 30 March 2026).

View source Report (multilateral / development partner)
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Pilot / Controlled Trial Phase
Year Initiated The year the AI system was first initiated or development began. 2018
Scale / Coverage The scale and geographic or population coverage of the deployment. Piloted in 12 comunas by October 2019; 2,262 children received intervention; planned rollout to all 345 comunas
Funding Source The source(s) of funding for the AI system development and deployment. Chilean national government budget (MDSF operational allocation)
Technical Partners External technology vendors, academic partners, or development partners involved. Not publicly disclosed
Outcomes / Results 2,262 children received intervention as of October 2019 across 12 pilot comunas. System recognised as a finalist in the 2019 Concurso Funciona government innovation competition. No published accuracy metrics or evaluation of predictive performance.
Challenges No public disclosure of algorithm type, model architecture, or accuracy metrics. Concerns about stigmatisation and labelling of children from lower-income communities. Data protection risks given doubly sensitive nature of children's personal data including health history and psychosocial information. Chile's data protection framework described by critics as inadequate. Lack of transparency about risk determination criteria and methodology. No citizen participation or consultation in system development.

How to Cite

DCI AI Hub (2026). 'Sistema de Alerta Ninez (SAN) -- ML-Based Child Welfare Risk Prediction', AI Hub AI Tracker, case CHL-002. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/CHL-002 [Accessed: 1 April 2026].

Change History

Updated 30 Mar 2026, 11:21
by system (system)
Created 30 Mar 2026, 11:18
by v2-import (import)