Artificial Intelligence (AI) applied to tax enforcement: an evaluation of the OSIRIS project in the face of the paradox of judicial inefficiency.

Authors

DOI:

https://doi.org/10.18256/2238-0604.2025.v21i3.5269

Keywords:

Tax Enforcement; Artificial Intelligence; OSIRIS Project; Judicial Efficiency; Technological Innovation.

Abstract

The Brazilian judicial system faces a major challenge with tax enforcement proceedings, characterized by a massive volume of lawsuits, high delays, and low efficiency in the recovering public credit. This article analyzes the dichotomy between the apparent simplicity of the TE legal procedure and the factual complexity faced in practice, based on data from the  report "Justice in Numbers 2024" (CNJ) and research on the cost of TE in the Federal District (FDRP-USP/IPEA). To address these challenges, the article examines the potential use of artificial intelligence through the OSIRIS Project, developed by the University of Brasília (UnB) in partnership with the Attorney General's Office of the Federal District (PGDF) and the Court of Justice of the Federal District and Territories (TJDFT). OSIRIS proposes an architecture guided by procedural logic, demonstrating high accuracy in identifying procedural phases, states, and acts, along with a significant reduction in computational costs. It is concluded that AI has great potential to optimize TE, reduce delays, and increase revenue collection, although further research is needed to precisely quantify these gains. The OSIRIS architecture also presents itself as a replicable model for other areas of law.

Downloads

Download data is not yet available.

Author Biographies

  • Debora Bonat, UnB

    Doutora em Direito pela Universidade de Brasília, Brasil, Professora do Programa e pós-graduação em Direito da UnB (Capes 7), Co-líder do Laboratório de Pesquisa DR. IA, Visiting Research Fellow- Université de Montréal. 
    E-mail: [email protected] | ORCID: https://orcid.org/0000-0002-0245-2380

  • Izabela Frota Melo, UnB

    Subprocuradora-Geral do Distrito Federal, Coordenadora do Projeto OSIRIS, Mestranda no Programa de Pós-Graduação da Faculdade de Direito da Universidade de Brasília, Brasil, Presidente da Rede Nacional de Governança, Estratégia e Inovação da Advocacia Pública Brasileira (RENAGEI). 
    E-mail: [email protected] | ORCID: https://orcid.org/0009-0002-6523-8848

References

BRAZ, Fabricio Ataides; SILVA, Nilton Correia da; NISHI, Luciana; SIQUEIRA, Eduardo Camargo de; BONAT, Debora; BERTONCINI, Cristina; MELO, Izabela Frota. OSIRIS: A Domain-Guided Framework for Tax Foreclosure Automation Using Large Language Models. In: INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND LAW, 2025, Chicago. Anais [ou Proceedings]. Nova Iorque: ACM, 2025. p. 1-5.

CONSELHO NACIONAL DE JUSTIÇA. Justiça em números 2024. Brasília: CNJ, 2024. Disponível em: bibliotecadigital.cnj.jus.br/xmlui/handle/123456789/860. Acesso em: 15 jul. 2025.

CORRÊA, Cristina Mendes Bertoncini; BONAT, Debora; ROCHA, Mariane Carolina Gomes da Silva; BRAZ, Fabricio Ataides; SILVA, Nilton Correia da; NISHI, Luciana; SIQUEIRA, Eduardo Camargo de. Os desafios no desenvolvimento de inteligência artificial aplicada ao direito: uma análise sobre o projeto Osíris. Revista Democracia Digital e Governo Eletrônico, Florianópolis, v. 1, n. 24, p. 56-69, 2025. Seção A. Edição Especial do 33º Encontro Ibero Americano de Governo Eletrônico e Inclusão Digital.

Documento de Oficialização de Demanda n. 25931498. In: PROCURADORIA-GERAL DO DISTRITO FEDERAL. Processo Administrativo SEI n. 00020-00029336/2019-06: Aprovação de Projeto. Brasília, 2019.

ESTEVES, Carolina Bonadiman. Mecanismos de citação do executado. In: CUNHA, Alexandre dos Santos; da SILVA Paulo Eduardo (Orgs.). Gestão e jurisdição : o caso da execução fiscal da União, v. 09, Brasília: IPEA, 2013.

FACULDADE DE DIREITO DE RIBEIRÃO PRETO DA UNIVERSIDADE DE SÃO PAULO (FDRP/USP). Projeto Custo Unitário da Execução Fiscal na Vara de Execuções Fiscais do Distrito Federal (VEF). Ribeirão Preto, SP: FDRP/USP, 2021. 21 p. Disponível em: https://www.pg.df.gov.br/wp-conteudo/uploads/2021/12/Relatorio_Final_Custo_EF_TJDFT_27.05.21__3_.pdf. Acesso em: 16 jul. 2025.

PÁDUA, Sérgio Rodrigo de; HARTMANN PEIXOTO, Fabiano. Inteligência Artificial Generativa no direito: da metodologia de avaliação à correção jurídica de LLMS e agentes de IA. Revista Brasileira de Direito, Passo Fundo, v. 21, n. 1, e5237, p. 1-38, jan.-abr. 2025. DOI: https://doi.org/10.18256/g4jnx825.

SARTOR, Giovanni. Artificial intelligence and human rights: Between law and ethics. Maastricht Journal of European and Comparative Law, v. 27, n. 6, p. 705-719, 2020. DOI: https://doi.org/10.1177/1023263X20981566.

ZUFELATO, Camilo. Em busca do tempo perdido no processo de execução fiscal médio: reflexões acerca da relação entre tempo útil e necessário e tempo morto. In: CUNHA, Alexandre dos Santos; da SILVA Paulo Eduardo (Orgs.). Gestão e jurisdição : o caso da execução fiscal da União, v. 09, Brasília: IPEA, 2013.

Downloads

Published

2026-08-10

Issue

Vol. 21 No. 3 (2025): setember-december

Section

Article

License

Copyright (c) 2026 Debora Bonat, Izabela Frota Melo

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

I declare the present article unpublished, as well as not be subject to any other submission process to another scientific journal.

I declare that I authorize the publication of the article by the magazine on the World Wide Web and the knowledge about non-remuneration by virtue of its publication, with no copyright of an equity nature.

Also, as author, I take civil and criminal responsibility for the content of the published work, after reading the guidelines for authors and agreeing with them.

The work can be accessed by any interested party and reproduced and / or published provided that proper reference is made according to the ABNT for the purpose of dissemination of scientific academic production.

How to Cite

BONAT, Debora; FROTA MELO, Izabela. Artificial Intelligence (AI) applied to tax enforcement: an evaluation of the OSIRIS project in the face of the paradox of judicial inefficiency. Revista Brasileira de Direito, Passo Fundo, RS, Brasil, v. 21, n. 3, p. e5269, 2026. DOI: 10.18256/2238-0604.2025.v21i3.5269. Disponível em: https://seer.atitus.edu.br/index.php/revistadedireito/article/view/5269. Acesso em: 23 aug. 2026.