Do banco de dados ao banco dos réus : reconhecimento facial e controle racial na Segurança Pública

Abstract

This study examines the use of facial recognition technology within the context of Brazilian public security, starting from the observation that biometric monitoring of faces has expanded in a continuous and diffuse manner, frequently dissociated from transparency, consent, and effective social oversight. The research demonstrates that such technology is already widely employed by governments and lawenforcement agencies on both a global and national scale, with particular emphasis on the significant expansion of initiatives in Brazil, including in small and mediumsized municipalities. The study analyzes the technical functioning of facial recognition technology, evidencing that its outputs do not constitute categorical identifications, but rather probabilistic statistical comparisons, inevitably marked by margins of error, especially in uncontrolled environments. The research further develops a critical assessment of the artificial intelligence systems underlying this technology, highlighting the presence of algorithmic bias, decision-making opacity, feedback loops, and the reproduction of structural inequalities, with a disproportionate impact on the Black population. It also examines the effects of facial recognition technology on fundamental rights and guarantees, such as privacy, the protection of personal data, individual liberty, the inadmissibility of unlawfully obtained evidence, and the presumption of innocence, as well as the absence of a specific and sufficiently robust regulatory framework within the Brazilian legal order. Finally, the study argues that the incorporation of facial recognition technology into public security practices, as currently implemented, intensifies penal selectivity and reinforces racialized surveillance practices, thereby assigning to Constitutional Law the role of a limiting framework for algorithm-mediated punitive power, in light of the principle of proportionality.