Artificial intelligence governance processes in Brazilian public sector organizations integrative review

Abstract

Artificial Intelligence (AI) has been spreading rapidly across multiple sectors and fields of knowledge, prompting debates about its potential benefits and risks. In the public sector, the adoption of AI-based solutions needs to ensure that such technologies are used ethically, responsibly, and in alignment with the collective interest. In this context, it becomes essential to discuss and structure governance mechanisms to guide these processes. The objective was to map governance processes proposed and adopted in public agencies in Brazil. To this end, an integrative literature review was conducted. The search was carried out in five databases—Scopus, Web of Science, Academic Search Premier (ASP), SciELO, and the Banco de Teses e Dissertações (BDTD)—covering the 2020–2025 period. A total of 400 documents were retrieved and, in the end, 13 comprised the analytic corpus: six master’s dissertations, two doctoral theses, and five journal articles. The studies were categorized into five theoretical axes of governance: ethical and regulatory aspects; risk and uncertainty management; transparency and fundamental rights; data governance and algorithmic auditing; and frameworks applied to specific contexts. It is concluded that AI governance in Brazil’s public sector is predominantly a field of propositions rather than consolidated practices. The initiatives identified are concentrated on ethical guidelines, risk frameworks, auditing mechanisms, and transparency recommendations. AI governance processes are conceived as institutional responses to regulatory and social pressures, but they still lack consolidation into tangible practices.