Mexico SAT — Machine Learning and Graph Analytics for High-Risk Taxpayer Detection (Plan Maestro 2024)

Mexico's Servicio de Administración Tributaria (SAT) has embedded machine-learning and graph analytics into its audit pipeline since 2020. Plan Maestro 2024 formalises a four-pillar strategy: Compliance Monitoring, Deep Surveillance, Coercive Collection, and Taxpayer Assistance.

Graph models map transactional networks among more than 80 million taxpayers in the CFDI database to expose concealed ownership and falsified invoice chains tied to Empresas Facturadoras de Operaciones Simuladas (EFOS, 'shell invoicers'). Statistical learning models target nine priority sectors — automotive, alcoholic beverages/cigarettes, construction, pharmaceuticals, hydrocarbons, logistics, technology platforms, real estate, and insurance/financial services — flagging inconsistencies between declared income, bank data, customs records and CFDI flows to generate risk scores for tiered interventions.

SAT reported collecting 4.517 trillion pesos in 2023, with a cost-to-collect ratio of 28 cents per 100 pesos — below the US IRS benchmark of 34 cents per dollar. The EFOS blacklist exceeded 14,000 entities by mid-2026, and 7,300 digital seals were suspended in the first half of 2026. New EFOS additions declined from more than 3,000 per year in 2018 to 47 in 2023.

Critics highlight risks of algorithmic bias, insufficient model transparency, and sector-specific scrutiny concentration. The decline in new EFOS additions is interpreted as either deterrence success or reduced enforcement, and no independent algorithmic audit has been published. Mexico's mandatory electronic invoicing, in place since 2014, provides an unusually rich machine-learning substrate among middle-income countries.