Algorithmic Explainability in Genomic AI: A Comparative Perspective with Proposals for Latin America

The convergence of artificial intelligence and genomics has generated systems for the automated processing of genetic information applicable to medical, insurance, and employment contexts, achieving levels of accuracy previously unattainable. This progress contrasts with the inherent opacity of algorithmic decision-making mechanisms, giving rise to conflicts between technological development and the protection of fundamental guarantees. This research conducts a critical examination of existing legal frameworks concerning algorithmic transparency applied to genomic artificial intelligence systems and reveals substantial limitations in addressing the distinctive characteristics of genetic data: its hereditary nature, relative genomic stability, predictive capacity, and social sensitivity. Through qualitative doctrinal analysis that combines comparative legal research with the study of landmark cases, the paper examines the regulatory frameworks of the General Data Protection Regulation (GDPR), the European Artificial Intelligence Act (AI Act), the Regulation on the European Health Data Space (EHDS), the Oviedo Convention, and other comparative regulatory developments. The study highlights crucial regulatory shortcomings: the lack of specific consideration of familial effects, the absence of adaptive temporal frameworks for genomic predictions, and the lack of principles for managing inherent scientific uncertainty. In response, it proposes “differentiated genomic explainability” as a new legal category incorporating familial contextualization, adaptive temporality, explicit uncertainty management, and distinctive proportionality. The findings underscore the urgent need to establish specific regulatory frameworks that reconcile the protection of fundamental rights with scientific advancement.