O uso da inteligência artificial pela Receita Federal do Brasil : a Seleção Aduaneira por Aprendizado de Máquina (SISAM) sob a ótica da proteção de dados

Abstract

This study analyzes the use of Artificial Intelligence (AI) by the Brazilian Federal Revenue Service (RFB), focusing on the Selection System for Customs Audit by Machine Learning (SISAM) from the perspective of data protection. The research investigates how the increasing automation of tax auditing activities, while improving tax oversight efficiency, may pose risks related to transparency, algorithmic opacity, and taxpayer discrimination. Considering the importance of the General Data Protection Law (LGPD), the study discusses the applicability of its principles to tax administration, highlighting the need for a balance between state efficiency and respect for citizen’s fundamental rights. The analysis also emphasizes the role of the National Data Protection Authority (ANPD) in auditing and regulating these technologies, given the limitation of data subject’s power to access information protected by trade secrecy or national security concerns. The study concludes that, although AI represents an advancement in tax auditing, its use must be guided by principles of algorithmic governance and data protection, ensuring that automation does not compromise constitutional rights.