인공지능 산출물과 학습데이터에 관한 저작권 문제 고찰

This article provides an in-depth analysis of two major issues:the copyrightability of AI-generated works and the potential copyright infringement during the AI training process. By comparing and analyzing case laws and legislation from various countries, as well as introducing diverse discussions from the academic community, this article offers a multifaceted examination of the relevant issues. Despite the fundamental changes in the creative environment brought about by advanced AI technologies, the current copyright law, designed with a focus on human creators, has limitations. The article suggests that the decision to grant copyright protection to AI-generated works should be based on whether the creativity of the work is sufficient to warrant protection as a copyrighted work. Through a thorough analysis of the AI Madam Suro case recognized as a compilation work by the Korea Copyright Commission, the U.S. Copyright Office’s decision to refuse registration of AI-generated works, the computer-generated works provision in the U.K. copyright law, and the Chinese court rulings protecting AI-generated works, this article identifies the positions of different countries and derives implications. Furthermore, the article analyzes the issue of copyright infringement during the AI training process. The collection and preprocessing of vast amounts of data are essential for AI training, and this process may lead to potential copyright infringement issues. There is a need to establish a reasonable scope of exemption for the use of works for deep learning purposes while ensuring that the rights of copyright holders are not excessively infringed upon during the data usage process. Under the current copyright law, copyright infringement can be established if the requirements of reliance and substantial similarity are met, but there is also a possibility of exemption under the fair use doctrine. Recently, numerous copyright infringement lawsuits have been filed against AI companies in the United States, drawing attention to this issue. Meanwhile, countries such as the United Kingdom, the European Union, and Japan have introduced text and data mining (TDM) exemption provisions, allowing the free use of copyrighted works under certain conditions. In South Korea, a proposed amendment to the Copyright Act, which includes the introduction of TDM exemption provisions, has been submitted. The copyright regulation of the AI training process should aim to harmoniously realize the two values of technological innovation and protection of creators.