Japanese Publishers Including Studio Ghibli Demand OpenAI Halt AI Training on Copyrighted Content

A coalition of Japanese publishers, led by the Content Overseas Distribution Association (CODA) and including Studio Ghibli, has formally demanded that OpenAI stop using their copyrighted works to train its AI models, particularly targeting the new Sora 2 video-generation tool. They argue that OpenAI's use of their content without permission violates Japan's copyright laws, which require prior authorization and do not permit liability avoidance through post-use objections. The publishers express concern that OpenAI's opt-out system for rights holders, where creators must actively request removal of their works, infringes on their rights and enables unauthorized replication of culturally significant content. OpenAI's previous approach of seeking forgiveness rather than permission has sparked similar controversies, as seen with ChatGPT’s image generation mimicking Studio Ghibli’s style, which drew widespread attention. CODA’s appeal extends to high-level Japanese officials to intervene and protect national creative assets, underscoring a broader tension between AI development and copyright enforcement. The dispute highlights the legal and ethical challenges of AI training data usage, with uncertain outcomes in courts due to outdated copyright laws and limited judicial precedent internationally.