COL-001

SISBEN IV -- AI-Enhanced Data Analytics and Fraud Detection Layer for Social Programme Targeting

Download PDF
Colombia Latin America & Caribbean Upper middle income Scaled & Institutionalised Confirmed

Departamento Nacional de Planeacion (DNP)

At a Glance

What it does Classification — Decision support for eligibility and benefits
Who runs it Departamento Nacional de Planeacion (DNP)
Programme SISBEN IV (Sistema de Identificacion de Potenciales Beneficiarios de Programas Sociales) -- Cross-cutting targeting system for 18+ social programmes including Familias en Accion
Confidence Confirmed
Deployment Status Scaled & Institutionalised
Key Risks Governance and institutional oversight risks
Key Outcomes 653,000 cases tagged 'under verification' through automated cross-referencing of pensions and health databases.
Source Quality 4 sources — Government website / press release, Academic journal article, Working paper / technical note, +1 more

Colombia's Sistema de Identificacion de Potenciales Beneficiarios de Programas Sociales (SISBEN), administered by the Departamento Nacional de Planeacion (DNP), is an algorithmic system that individually rates the Colombian population according to socioeconomic status to determine eligibility for social programmes. SISBEN has operated since 1994 and has undergone four major revisions. The fourth version (SISBEN IV) introduced significant changes including automated cross-referencing of 34 or more administrative databases, data analytics technologies, and automated anomaly detection to identify inconsistencies and potential fraud in beneficiary data.

SISBEN produces an individual score from 0 to 100 (where 100 indicates greater prosperity) and classifies people into groups based on living conditions and income. Each social programme entity sets its own cut-off points on the SISBEN score or group to determine eligibility. The system is used for targeting at least 18 social programmes of different characteristics, and the subsidised health system using SISBEN covers more than half of Colombia's population. Plans project expansion to reach 40.5 million people, approximately 84 percent of the population.

The SISBEN IV redesign was motivated by a 2016 analysis conducted jointly by DNP, the World Bank, and ECLAC that identified critical problems with SISBEN III: the absence of an income component, the absence of an interoperable system to verify citizen-reported information, and an algorithm that concentrated 50 percent of the score on health, education, and housing variables. SISBEN IV shifted from a quality-of-life framework to a 'presumption of income' and 'income generating capacity' approach, and introduced interoperability with administrative databases covering health, pensions, education, work, real estate, taxes, financial risks, social benefits, transportation, victim registration, and public services.

The automated fraud detection and validation component is a key innovation. Decree 441 of May 2017 gave DNP charge of database validation and quality controls and enabled public entities to share information without formal agreements. Cross-referencing of pensions and health system databases with the SISBEN database identified 653,000 cases tagged 'under verification' for high income discrepancies and deceased persons. There are nine defined reasons for tagging cases as 'under verification,' ranging from unreported changes of residence to income records higher than DNP-determined thresholds. Upon tagging, territorial entities inform the affected person and decide on exclusion or reclassification; within six months, benefit-administering entities are notified to withdraw benefits if warranted.

The system has attracted significant legal and civil society scrutiny. Colombia's Constitutional Court, in Ruling T-716/17, found the system's design to be 'arbitrary or unfair' while appearing to be objective. The Karisma Foundation, a Colombian civil society organisation, filed access-to-information requests about SISBEN's algorithm, which DNP refused citing confidentiality because revealing the algorithm 'may compromise the country's macroeconomic and financial stability' and 'constitutes fraud.' Academic analysis by Joan Lopez (Tilburg University) and Juan Diego Castaneda, published through the Karisma Foundation in 2020, characterised SISBEN as an 'algorithmic assembly' that creates a surveillance-like integration of citizen data across State institutions, noting that citizens classified by the system cannot know how their data is classified and lack means to demand explanations for their rating.

Documented manipulation includes local officials modifying scores so people can receive benefits, threats to people in the SISBEN database to vote for certain candidates in exchange for keeping benefits, and local governments inflating low-scoring populations to receive bigger budgets. A 1997 assessment had already found increasing fraud in how SISBEN scores were obtained.

No specific algorithm type, model architecture, or technology vendor has been publicly disclosed. DNP develops specialised software internally that generates individual scores and structures the population classification.

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

Social Protection Functions

Implementation/delivery chain
Assessment of needs/conditions + enrolment primaryRegistration
Programme design
Eligibility criteria and qualifying conditions
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name SISBEN IV (Sistema de Identificacion de Potenciales Beneficiarios de Programas Sociales) -- Cross-cutting targeting system for 18+ social programmes including Familias en Accion
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Poverty targeted Cash Transfers (conditional or unconditional)
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 Algorithmic poverty classification and fraud detection system that individually rates Colombia's population by socioeconomic status using statistical models and automated cross-referencing of 34+ administrative databases, used to determine eligibility for at least 18 social programmes.
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 Monitoring, Maintenance and Decommissioning
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 Developed in-house
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 I — Sovereign AI Zone
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Domestic
Data Residency Detail Additional detail on the specific data hosting arrangements and jurisdictions. DNP is a national government entity; SISBEN database maintained domestically with data from Colombian government agencies
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary None
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 HOTL
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 Fully in-house
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. Formal assessment
Documented Risk Events Constitutional Court Ruling T-716/17 found the system 'arbitrary or unfair.' DNP refused access-to-information requests about the algorithm citing macroeconomic stability and fraud concerns. Local officials documented modifying scores for political manipulation. Threats to beneficiaries to vote for candidates in exchange for keeping benefits. Local governments inflating low-scoring populations for budget manipulation.

Impact Dimensions

Equality, non-discrimination, fairness and inclusion
  • Grievance mechanism
  • Independent evaluation
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsSensitiveLinks data across multiple systemsCurrently available and usedHealth records, pension records, education records, employment records, real estate records, tax records, financial risk records, and other government databases cross-referenced for validation and fraud detection.
Social registriesSensitiveLinks data across multiple systemsCurrently available and usedHousehold survey data collected door-to-door by municipalities using DNP-developed software, cross-referenced with 34+ administrative databases covering health, pensions, education, employment, real estate, taxes, financial risks, social benefits, transportation, victim registration, and public services.

DNP (n.d.) 'Que es el SISBEN?', Departamento Nacional de Planeacion. Available at: https://www.sisben.gov.co/Paginas/que-es-sisben.html (Accessed: 30 March 2026).

View source Government website / press release

Lopez, J. and Castaneda, J.D. (2020) 'Automation, Digital Technologies and Social Justice: The Case of SISBEN in Colombia', in Cerrillo-i-Martinez, A. and Fabra-Abat, P. (eds.) AI, Human Rights and Social Justice. Karisma Foundation / CETyS Universidad de San Andres. Available at: https://web.karisma.org.co/ (Accessed: 30 March 2026).

View source Academic journal article

Farhat, Y. (2021) 'Chosen by a Secret Algorithm: Colombia's top-down pandemic payments', Center for Human Rights and Global Justice, NYU School of Law. Available at: https://chrgj.org/2021-12-14-transformer-states-colombia/ (Accessed: 30 March 2026).

View source Working paper / technical note

fAIrLAC / IDB (n.d.) 'SISBEN', fAIrLAC, Inter-American Development Bank. Available at: https://fairlac.iadb.org/en/sisben (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 Scaled & Institutionalised
Year Initiated The year the AI system was first initiated or development began. 1994
Scale / Coverage The scale and geographic or population coverage of the deployment. Used for 18+ social programmes; subsidised health system covers more than half the population; plans to reach 40.5 million people (84% of population); 653,000 cases tagged 'under verification' through automated cross-referencing
Funding Source The source(s) of funding for the AI system development and deployment. Colombian national government budget (DNP operational allocation)
Technical Partners External technology vendors, academic partners, or development partners involved. World Bank and ECLAC (supported SISBEN III analysis and IV redesign methodology)
Outcomes / Results 653,000 cases tagged 'under verification' through automated cross-referencing of pensions and health databases. System used to target at least 18 social programmes. Subsidised health system using SISBEN covers more than half the population. Constitutional Court scrutiny has led to ongoing legal and governance reforms.
Challenges Algorithm and variable specifics not publicly disclosed (DNP claims confidentiality). Integration of 34+ databases creates surveillance risks. Socioeconomic manipulation by local officials documented. Citizens lack right to explanation of their classification. Political malleability since DNP leadership is appointed by the President. Exclusion errors from tightening requirements not well analysed. System presented as error-proof despite documented quality and handling failures.

How to Cite

DCI AI Hub (2026). 'SISBEN IV -- AI-Enhanced Data Analytics and Fraud Detection Layer for Social Programme Targeting', AI Hub AI Tracker, case COL-001. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/COL-001 [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)