POLICY ON THE FORMULATION OF ARTIFICIAL INTELLIGENCE ACCOUNTABILITY FOR THE CREATION AND DISSEMINATION OF DEEPFAKE VIDEOS FROM A CRIMINAL LAW PERSPECTIVE IN INDONESIA

Abstract

The advancement of Generative Adversarial Network (GAN) technology has enabled deepfakes to facilitate serious crimes in Indonesia, including non-consensual pornography, public figure impersonation, and political disinformation. This study analyzes Indonesia's existing criminal law framework regarding AI-based deepfake accountability and proposes ideal policy reforms. Using normative legal research with legislative, conceptual, and comparative approaches, this study identifies significant normative gaps (lex lacuna) across the ITE Law, the New Criminal Code or Kitab Undang-Undang Hukum Pidana (KUHP), and the Personal Data Protection (PDP) Law. Key deficiencies include the absence of a legal definition of deepfakes, no comprehensive accountability mechanism covering the AI production chain, and difficulties applying mens rea principles given AI system autonomy. Comparative analysis of the EU AI Act and China's Deep Synthesis Regulations reveals that effective regulation requires a layered accountability model: strict liability for AI developers, vicarious liability for platform operators, and direct liability for end users. This framework should be complemented by mandatory synthetic content labeling and harmonization with the PDP Law and New Criminal Code as lex generalis. This study recommends either enacting a dedicated Artificial Intelligence Bill or revising the ITE Law to explicitly criminalize deepfake creation and dissemination. Such reform is essential to establish legal certainty, close existing normative gaps, and strengthen public protection against AI-facilitated crimes in Indonesia.