SAT's AI-Driven Tax Fraud Detection and OrientaSAT Chatbot — Mexico

Since October 2017, Mexico's Servicio de Administración Tributaria (SAT) has applied machine-learning models to its mandatory nationwide electronic-invoicing system (CFDI), comparing declared income and VAT bases against issued invoices to flag simulated transactions and shell companies ("empresas fantasma"). In 2020 the models were credited with identifying roughly 1,200 shell companies and about 3,500 fraudulent transactions.
In March 2021, SAT added OrientaSAT, a free AI chatbot that assists taxpayers during filing season; its interactions grew 47.5% to 246,309 in 2023. By Q1 2024, SAT reported a 25.4% increase in audit efficiency and a recovery ratio of 201.6 pesos per peso invested in enforcement, versus 160.8 pesos a year earlier, attributing part of the gain to AI-assisted risk classification under its 2024 "Plan Maestro."
A peer-reviewed study by researchers affiliated with UNAM raises concerns about algorithmic bias against specific taxpayer groups, the non-disclosure of the risk-scoring methodology, and unclear legal accountability; it also notes that SAT's headcount fell by 7,004 positions (-22%) in 2022, the sharpest reduction on record, coinciding with the automation push. SAT's own causal claims about revenue gains from AI specifically, as opposed to broader digitalisation, have not been independently audited.

Read the full analysis: https://www.sat.gob.mx/