Open-Source AI and Legal Integrity in Latin American Courts: Governing Algorithmic Justice in Brazil and Beyond
Year: 2025
Type: conference-paper
Abstract: Open technologies are crucial for transparency, resilience, and accountability, particularly in the face of rapid technological advancements and institutional vulnerabilities. This significance intensifies in politically and economically uncertain contexts. Focusing regionally on Latin America, this study emphasizes Brazil’s judiciary as a regional leader in integrating artificial intelligence (AI). In Brazil, key AI initiatives, including CNJ Resolution No. 332/2020, mandate transparency, fairness, explainability, and the preservation of human oversight. National platforms such as Sinapses and Codex/DataJud exemplify collaborative and standardized approaches. However, proprietary solutions like the Supreme Federal Court’s VICTOR underscore ongoing challenges concerning opacity, limited auditability, and due process. This paper provides a comparative analysis of other Latin American countries, revealing varying transparency levels and governance frameworks. Argentina’s Prometea and AymurAI platforms are open-source, emphasizing rights-based frameworks. In contrast, Colombia’s mixed approach with tools like Copilot and ChatGPT reflects varying transparency standards, while Mexico’s JulIA initiative indicates limited accountability. Identified challenges across these contexts include algorithmic biases, inadequate oversight mechanisms, and regulatory gaps in data protection, particularly within Brazil’s General Data Protection Law (LGPD) and Mexico’s LFPDPPP. The paper proposes solutions to enhance AI governance: mandating open source development for publicly funded AI projects, implementing independent algorithmic audits, ensuring explainability through comprehensive documentation, and offering AI ethics education specifically tailored for judicial personnel.
Author: Maria Lombardi, Ettore
Institution: University of Florence
Language: English
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7105662095
FWCI:
Cites:
Cited by:
Related to: