Research on the Adjudication of Infringement in the Use of Works in AI Data Training and the Statutory License System

In response to the national policy directive on improving the development and governance mechanisms for generative artificial intelligence, and to address the challenges posed by AI technology to the current copyright system, this paper focuses on the core legal issues surrounding the use of works in AI model data training. The research first deeply analyzes the nature of the use of works during the data collection and model training stages, demonstrating that constitutes acts of use regulated under the current copyright law. Thus, unauthorized use constitutes an infringement upon the exclusive rights of copyright holders, such as the right of reproduction. This paper, grounded in an analysis of the operational mechanisms of data training in artificial intelligence, argues that the use of copyrighted works during the stages of data collection and model training constitutes a form of use subject to regulation under existing copyright law. Accordingly, the unauthorized use of such works by AI technology companies for training purposes, absent the consent of the rights holders, shall be deemed an act of infringement. This paper contends that neither the traditional licensing model nor the proposed fair use approach can adequately resolve the tension between the protection of copyright holders’ rights and the advancement of artificial intelligence technologies. An overly stringent intellectual property regime may undermine the interests of rights holders, while an excessively permissive framework risks stifling technological innovation. To address this dilemma, the paper advocates for the appliance of statutory license to the field of data training. Such a regime would enable AI technology companies to make large-scale use of copyrighted works while ensuring that the legitimate interests of rights holders are duly respected. Besides, this approach aligns with the dual value orientation of copyright law, which seeks to balance protection with limitation.