Artificial intelligence in the Brazilian federal revenue service: automation of risk management, detection of tax fraud, and algorithmic governance

Abstract

The digital transformation of the public sector has expanded the use of data analysis tools, automation, and artificial intelligence in administrative and oversight activities. In the case of tax administration, this movement responds to the growing volume of fiscal data, the complexity of digital economic transactions, and the need to improve risk management. This study analyzes how Brazil's Federal Revenue Service has been incorporating artificial intelligence resources to support risk management, the detection of tax and customs fraud, and the structuring of algorithmic governance mechanisms. Methodologically, this is a bibliographic and documentary research study with a qualitative approach, based primarily on official institutional documents, regulatory acts, and official institutional communications made available by the Federal Revenue Service, the Ministry of Finance, Serpro, and public government repositories, complemented by academic literature on digital transformation and artificial intelligence in public administration. The results indicate that the initiatives analyzed are concentrated in analytics platforms, complex network analysis, cryptoasset monitoring, risk pattern detection, and the institutionalization of principles of accountability, transparency, and human oversight. It is concluded that the Federal Revenue Service's experience reveals a data-driven modernization process in which artificial intelligence ceases to be merely an operational instrument and becomes integrated into the state's capacity for oversight and decision-making, while simultaneously requiring governance safeguards, regulatory compliance, and human control.