AI 기반 온라인 게임의 저작권 침해 판단 기준에 관한 연구

The recent rapid development of generative AI technology is bringing about innovative changes to the entire content production industry, including the online game industry. AI is playing a key role in the entire game development process, and in particular, it is drastically reducing development time and cost through automatic creation of content such as character creation, scenario composition, and world view design, while increasing the depth and immersion of the user experience. Furthermore, the introduction of real-time interactive content and 3D digital actor technology is further enhancing the realism and immersion of game content. However, such technological advancements are accompanied by new legal issues, especially multi-layered issues related to copyright infringement. Since generative AI can automatically generate similar content by learning the structure, style, and narrative of existing works, the criteria for determining whether there is copyright infringement are becoming ambiguous, and there are limitations in sufficiently regulating this with the existing legal system alone. In fact, numerous game copyright dispute cases, such as NCsoft vs. Webzen and Nexon vs. Iron Mace, show that AI technology is increasing the similarity between games and thus increasing the possibility of legal disputes. Accordingly, this paper examines the necessity of establishing criteria for copyright infringement due to the introduction of generative AI in game content production, and examines major issues from various angles, such as the creative contribution of the prompt inputter, the legal responsibility of the algorithm designer, and the criteria for judging the similarity of the product. In particular, prompts can be considered creative acts that can include narrative structure, emotional flow, and style design beyond simple commands, and if a certain level of creativity is recognized, the inputter can also be granted legal status under copyright. On the other hand, prompts that induce imitation of others' styles or narratives can be interpreted as indirect infringement or inducing infringement, so legal criteria for judgment should also be established. In addition, securing transparency in the composition of the AI ​​algorithm's learning dataset, structural limitations in the iterative generation process, and whether real-time generated content is jointly created are also important factors to consider in copyright judgment. Therefore, based on the various legal and technical issues discussed in this paper, it is necessary to establish new criteria and an interpretation system that can more precisely judge whether there is copyright infringement in generative AI-based game content.