ARTIFICIAL INTELLIGENCE AND DIGITAL SEXUAL VIOLENCE

DEEPFAKES, FAKE NUDES, AND INSTITUTIONAL RESPONSES IN BRAZIL

Abstract

The fast spread of generative AI embedded in digital platforms has reshaped technology-facilitated sexual violence by enabling the large-scale creation and circulation of synthetic intimate content, including sexual deepfakes and “fake nudes”. This article argues that AI should be understood not merely as a technical tool but as a social institution: it structures practices, incentives, and risks, producing foreseeable and systemic harms, disproportionately affecting women and girls. From this institutional lens, the paper highlights the limits of traditional legal categories—especially consent, which presupposes a free, prior human manifestation of will—an element absent in algorithmic manipulation of images and videos. The analysis then examines shortcomings of Brazilian legal responses, focusing on constitutional protections of dignity, honor, and image; the liability framework under the Marco Civil da Internet; Brazil’s data protection law (LGPD); recent criminal law amendments; and Law No. 15.211/2025 (Digital ECA), emphasizing regulatory gaps, evidentiary challenges, and the inadequacy of purely reactive enforcement. Finally, the article advances an institutional-arrangements approach, calling for preventive and ongoing duties of due diligence, risk-mitigation “by design”, coordinated institutional action, and public policies capable of addressing the massive, transnational, and automated nature of digital sexual violence.