THE (IN)SECURITY OF EVIDENCE IN THE ERA OF ARTIFICIAL INTELLIGENCE: CHALLENGES TO EVIDENTIARY VERACITY IN THE PROTECTION OF CHILDREN'S PERSONALITY RIGHTS

Abstract

This article analyzes the risks that generative artificial intelligence poses to evidentiary veracity and to the effective protection of children's personality rights in the Brazilian legal system. It starts from the premise that the relative presumption of veracity historically attributed to digital evidence rests on technical assumptions now overcome by the sophistication of deepfakes, synthetic content technically indistinguishable from reality. Adopting a qualitative, juridical-dogmatic methodology based on bibliographical and documentary research, the study reveals a twofold vulnerability: Brazilian criminal legislation presupposes a real photographed victim, creating a gap in typification regarding synthetically generated Child Sexual Abuse Material (CSAM); meanwhile, forensic fragility fosters both wrongful incrimination and the so-called "liar's dividend", whereby genuine evidence is discredited. It is concluded that a multi-level normative and institutional framework is required — combining AI regulation, autonomous criminal typification, synthetic media forensics and international cooperation — to ensure the effectiveness of the Doctrine of Integral Protection in the digital era.