Japan’s Content Overseas Distribution Association (CODA), representing major intellectual property holders including Studio Ghibli and Bandai Namco, has formally urged OpenAI to cease using its members’ copyrighted content to train the Sora video generation model without permission. The request comes amid growing backlash over AI-generated content that mimics the distinctive artistic styles of Japanese creators, particularly following the launch of Sora 2 on September 30. The app quickly produced a wave of videos featuring characters and visuals resembling those from Studio Ghibli films, including “Spirited Away” and “My Neighbor Totoro,” sparking outrage among artists and rights holders. The issue is not new. When OpenAI’s image generator was introduced in March, users began creating “Ghiblified” versions of selfies, pets, and even celebrity portraits, with OpenAI CEO Sam Altman famously changing his X profile picture to a stylized version of himself in a Ghibli aesthetic. The trend highlighted how easily generative AI can replicate and reproduce copyrighted artistic styles, raising concerns about unauthorized use and potential devaluation of original creative work. CODA argues that the process of training AI models on copyrighted material—especially when the output closely imitates protected works—constitutes copyright infringement under Japanese law. Unlike U.S. copyright law, which allows for some flexibility through the “fair use” doctrine, Japan’s system generally requires prior permission before using copyrighted content. CODA emphasized that Japan does not recognize a legal framework where companies can claim immunity by offering an opt-out after the fact. “Under Japan’s copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections,” the group stated. The situation has drawn attention from Japan’s government, which has formally requested OpenAI halt the replication of Japanese artwork. This follows a pattern of similar concerns from other rights holders, including Nintendo and the estate of Dr. Martin Luther King Jr., who have also raised alarms about the potential for deepfakes and unauthorized use of their intellectual property through AI tools like Sora. While OpenAI has announced plans to update Sora’s opt-out policy, allowing content owners to request removal of their works from training data, CODA argues this is insufficient. The association maintains that using copyrighted material without consent—even if later removed—may already constitute infringement under Japanese law. The opt-out model, they say, fails to meet the requirement for prior authorization. Hayao Miyazaki, Studio Ghibli’s legendary co-founder and creative force behind its iconic films, has not publicly commented on the AI trend, but in 2016 he expressed strong disapproval of AI-generated animation, calling it “utterly disgusting” and an “insult to life itself.” His sentiments reflect broader concerns within the creative community about the ethical and artistic implications of AI replicating human creativity. This case underscores the global divide in how copyright law applies to AI training. While a recent U.S. ruling by Judge William Alsup found that Anthropic’s use of legally purchased books for training was fair use, the decision did not extend to pirated content or address output-based infringement. Japan’s stricter stance suggests that future legal battles may hinge on jurisdiction, with rights holders in countries like Japan potentially having stronger grounds to sue. For now, CODA is calling on OpenAI to respond sincerely to copyright claims and halt the use of Japanese IP in training data. The outcome could shape how AI companies operate internationally and whether they must obtain explicit permission before using copyrighted content—setting a precedent that may influence global AI development.