Anthropology of algorithmic regulation

disputes over AI regulation in Bill no. 2338/2023

Abstract

This article proposes an anthropological approach to algorithmic regulation through an analysis of Bill No. 2338/2023, the most consolidated legislative initiative aimed at establishing a regulatory framework for artificial intelligence in Brazil. It examines how categories such as risk, rights, governance, and social participation are disputed in the construction of the Brazilian regulatory design. Methodologically, it combines documentary analysis and the mapping of regulatory controversies, drawing on versions of the bill, amendments, reports, legal opinions, technical notes, public hearings, and documents produced by governmental actors, companies, civil society organizations, and specialists. The analysis argues that AI governance in Brazil is organized around three overlapping layers: AI as innovation, AI as a public problem, and AI as infrastructure. This classification helps explain how regulatory models formulated in other contexts should be examined in light of the institutional, political, and economic asymmetries affecting countries in the Global South. The article offers a situated understanding of AI regulation, emphasizing the entanglement of legal norms, infrastructures, and power relations.