LGPD, inteligência artificial e compliance: o dever da transparência algorítmica

Abstract

The growing delegation of decision-making to artificial intelligence systems, fueled by large volumes of personal data, has established a structural tension between the opacity of these Technologies, the so-called "black box" phenomenon, and the effectiveness of data subjects rights. Against this backdrop, this study investigates how the Brazilian General Data Protection Law (Law No. 13,709/2018) may guide and require compliance practices aimed at algorithmic transparency, reconciling the protection of fundamental rights with the preservation of business competitiveness and innovation. A qualitative approach was adopted, based on bibliographic review and documentary analysis, guided by the legal-dogmatic method. The research demonstrates that algorithmic opacity undermines rights such as privacy, equality, nondiscrimination, and informational self-determination, and that the LGPD provides the normative foundations for algorithmic transparency by enshrining the principles of transparency and accountability, guaranteeing the right to explanation and to the review of automated decisions, and establishing the data protection impact assessment. It is acknowledged, however, that absolute transparency is unattainable in light of technical, legal, and practical constraints, which is why this study advocates for proportional transparency, calibrated according to the risk and impact of each system. It is concluded that algorithmic transparency and compliance are complementary facets of a single requirement, as the law provides the normative foundation and compliance provides the means for its practical implementation, institutionalizing intelligibility, auditability, and accountability.