COPYRIGHT PROTECTION AGAINST THE USE OF WORKS AS ARTIFICIAL INTELLIGENCE TRAINING DATA IN INDONESIA

Abstract

The advancement of generative artificial intelligence has created new legal challenges in the domain of intellectual property rights, particularly copyright law. The training of AI models requires the use of massive amounts of copyrighted works as training data; however, Indonesian Law Number 28 of 2014 on Copyright does not explicitly regulate the legality of such practice. This research aims to analyze the legal standing of the use of copyrighted works as AI training data under the Indonesian copyright law, to identify existing normative gaps, and to formulate an ideal licensing model for AI development in Indonesia. Employing normative legal research through statute, conceptual, and comparative approaches, this research examines the regulatory frameworks of the United States, the European Union, Japan, and the United Kingdom as comparative material. The findings indicate that the use of copyrighted works as AI training data potentially infringes the creator’s exclusive rights, particularly the right of reproduction, while no exception or limitation in Indonesia’s Copyright Law specifically accommodates such use. This research proposes an AI training-data licensing model built on a collective-licensing mechanism, accompanied by a national opt-out scheme and dataset-transparency obligations, as a normative solution that balances copyright protection with the need for artificial intelligence innovation in Indonesia.