생성형 AI 학습과 확대된 집중관리제도 - 웨스트로 판결 및 클로드 판결을 중심으로 -

Copyright, as its name suggests, treats the right to copy as its most central tenet. The concept of copying originated from the traditional idea of permanently fixing a work onto a physical medium, but it has expanded in the digital age to include even the temporary fixation of data in RAM. Under the modern concept of copying, when AI crawls, exploring web pages to extract data and materials in the digital space for learning purposes, it also constitutes a reproduction of a copyrighted work, thus raising the potential for copyright infringement. Indeed, copyright holders, facing the loss of market competitiveness and infringement of their rights due to the development of AI technology, are filing lawsuits against AI companies for copyright infringement. AI companies, in return, are trying to justify AI training as fair use. In response, U.S. courts have provided authoritative interpretations on AI training and copyright infringement through the Westlaw ruling and the Claude ruling. Although the conclusions of these two rulings differ, their implications for copyright policy concerning AI point towards “fair compensation for the use of copyrighted works.” Therefore, to create an environment where AI companies can use copyrighted works by paying fair compensation to copyright holders, negotiations by a collective management organization, such as the Korea Music Copyright Association (KOMCA) in Korea, are necessary. Furthermore, it is believed that a move towards an extended collective licensing system is needed, where the effects of negotiations extend even to non-entrusted rightsholders.