The Content Overseas Distribution Association (CODA), a prominent Japanese organization dedicated to safeguarding intellectual property rights and combating digital piracy on behalf of major creative companies such as Studio Ghibli and Bandai Namco, issued a formal letter last week calling on OpenAI to immediately cease the use of its members’ creative materials in the training of its Sora 2 artificial intelligence model. As reported by Automaton, this appeal reflects a growing unease among Japan’s entertainment and media sectors about the unlicensed utilization of copyrighted content for machine learning purposes.
In its letter, CODA asserts that the act of copying or replicating creative works during the AI training process might itself constitute a direct infringement of copyright. The organization argues that even if such duplication occurs indirectly or merely as part of data ingestion, it falls within the scope of legal protection granted to original works. This concern has deepened because the final product—Sora 2—subsequently generated an extensive amount of material featuring recognizable characters, imagery, and stylistic elements clearly derived from copyrighted Japanese intellectual properties. CODA’s position suggests that the reproduction of distinctive creative expressions, even by algorithmic or automated systems, cannot be excused under current copyright frameworks.
Following Sora 2’s public release on September 30, an overwhelming wave of AI-generated content appeared online, almost immediately displaying visual and narrative motifs reminiscent of well-known Japanese franchises. This surge prompted the Japanese government itself to formally demand that OpenAI refrain from further replicating any works of domestic origin, particularly in art and animation styles cherished around the world. Notably, this was not the first controversy of its kind: earlier in March, the debut of OpenAI’s GPT-4o model had resulted in a flood of so-called “Ghibli-style” illustrations circulating across social media—images that drew unmistakable inspiration from the aesthetic hallmarks of Japan’s beloved animation houses. The influence was so pronounced that even OpenAI’s CEO, Sam Altman, continues to use a profile portrait on X that strongly evokes the visual spirit of Studio Ghibli’s films.
In response to growing public pressure and criticism from rights holders, Altman announced last month that the company intends to revise Sora’s existing opt-out policy, which determines how intellectual property owners can exclude their materials from AI training datasets. Nevertheless, CODA contends that the very existence of an opt-out mechanism is legally problematic under Japanese copyright law. The association notes that, within Japan’s legal framework, creators must grant permission before a copyrighted work is used; there is no established provision allowing a company to proceed without prior consent and later claim exemption from liability through retroactive objections. In essence, CODA’s argument underscores a fundamental divergence between consent-based copyright regimes and the data-centric practices currently employed by large AI firms.
Representing the collective voice of its member organizations, CODA has now formally requested that OpenAI respond in good faith to each member’s copyright claims and immediately suspend the use of their creative works within any machine learning process, including both the training of future models and the generation of derivative outputs such as those produced by Sora. The organization’s demand extends beyond visible AI-generated content to the underlying training data itself, emphasizing that unapproved extraction or replication of Japanese intellectual property for algorithmic purposes undermines the principles of authorial control, fair use, and respect for cultural art forms. By pressing these concerns, CODA highlights an increasingly global challenge: how to reconcile rapid advances in generative AI with long-established norms of creative ownership and artistic integrity.
Sourse: https://www.theverge.com/news/812545/coda-studio-ghibli-sora-2-copyright-infringement
