Regulação da Inteligência Artificial na Administração Pública brasileira

Abstract

This study analyzes the regulation of Artificial Intelligence (AI) in the Brazilian Public Administration, investigating how to harmonize the principle of efficiency with other constitutional precepts and the fundamental rights of citizens. The research explores the Brazilian normative landscape, focusing on Bill of Law No. 2.338/2023, and contrasts it with referential frameworks, such as the European Union's AI Act. Using the deductive method and the bibliographic review methodology, this research proposes an adaptive governance model, based on three hypotheses: the adoption of a risk-based "framework law"; the complementarity between state heteroregulation and the need for public self-regulation; and the institutionalization of an Administrative Law of Experimentation through regulatory sandboxes to foster learning and more effective technical standards. It is concluded that, although Bill of Law No. 2.338/2023 represents a significant advancement, its refinement is necessary to avoid stifling innovation and to ensure the effectiveness of control mechanisms, such as human oversight and the explainability of acts generated by AI techniques. This will ensure that the technology serves as a tool to support the public interest, always subordinated to the dignity of the human person.