Sem fair play aos direitos fundamentais quando o reconhecimento facial entra em campo: um olhar crítico sobre o uso da tecnologia, como parte da política criminal estatal, nos estádios brasileiros

Abstract

Despite the discourse of modernization that accompanies facial recognition technologies, their implementation in Brazilian soccer stadiums reveals an old logic: that of surveillance as a technique of government. As Nilo Batista observes in his reading of Dom Casmurro, the Rio de Janeiro Correctional Facility in 1850 already expressed the transition from the slave-based penal system to dependent industrial capitalism, inspired by Benthamite panopticism. Although the project failed, the intention to meticulously control bodies and behavior was clear. This vigilant rationality is updated in the present in electronic and algorithmic form, legitimized by narratives that, as Ellen Rodrigues points out, use the myth of escalating violence to naturalize surveillance practices and increased policing. This rationality is modernized today in electronic vigilantism, in which cameras and algorithms take on the role of the “panoptic eye” in soccer stadiums—notably arenas such as Maracanã, Allianz Parque, Mineirão, Castelão, and Batistão. The problem examined is the tension between the promise of security and efficiency in managing large crowds and the risks of violating fundamental rights, notably the right to privacy and non-discrimination. The hypothesis formulated is that facial recognition, although presented as a neutral tool, operates selectively, exposing vulnerable groups to greater risks of criminalization and misidentification. The theoretical framework adopted articulates Zuboff's critiques of surveillance capitalism, Beck's critiques of risk society, and Casara and Dardot & Laval's critiques of neoliberal rationality, adding to Doneda's warnings about the fragility of personal data protection and the Brazilian legal framework for transparency and data protection (CF/88, LAI, and LGPD, especially Articles 6 and 11). On a technical level, we analyze the foundations of artificial intelligence applied to facial recognition, from the first Eigenfaces algorithms to the current predominance of deep neural networks and vision transformers. Although they have relevant accuracy rates, their operation in real events requires robust governance, including human double-checking, audit trails, and explicit quality parameters, and such systems remain vulnerable to algorithmic biases related to race, gender, and contextual conditions, which increases the risk of errors in mass surveillance environments such as stadiums. Methodologically, the research combines a critical literature review with empirical data collected through requests for information sent to different states of the federation (AL, BA, CE, SC, PA, PE, PB, PR, SE, RJ, SP, MG, and DF). To date, there have been partial and heterogeneous responses—with gaps in documentation, discrepancies between agencies, and responses that mention “tests” or “proofs of concept” without proper evidence of legal instruments, legal bases, and disposal procedures. Even when labeled as tests, such operations require some form of agreement/contract, RIPD, and data retention/disposal policy. The central empirical focus shifts to the cases of Castelão (CE) and Batistão (SE), in which evidence of operational and governance failures — including mistaken alerts with potential wrongful arrests — make the asymmetry between the promise of efficiency and the risks of rights violations more visible. In conclusion, it is demonstrated that facial recognition in stadiums is not merely a technological advance for the comfort of fans, but a device of power aligned with surveillance capitalism. Its legal legitimation, without adequate democratic safeguards, deepens inequalities and weakens fundamental rights, imposing the need for effective regulation with minimum guidelines and critical reflection on the limits of social control in neoliberal societies.