Japan’s biggest entertainment companies are pushing back against OpenAI’s video AI model Sora 2, filing a formal request to stop training it on Japanese works without permission.
As reported by Automaton, industry group CODA has issued a written request asking OpenAI to end the unauthorized use of Japanese anime, manga, and game content in Sora 2’s development.
CODA represents leading rights holders including Aniplex, Bandai Namco, Studio Ghibli, Square Enix, and Kadokawa. The group says Sora 2’s visuals strongly resemble existing Japanese works, indicating that those materials were used without approval during training.
Japanese publishers demand OpenAI stop using their content
In the October 28 request, CODA argues that Japanese copyright law requires creators to grant permission before their works are used for machine learning. It also rejects opt-out systems, stating that companies cannot avoid infringement claims by removing content only after a complaint has been filed.
CODA has also asked OpenAI to respond clearly to any future copyright inquiries from member publishers. On October 2, government officials warned OpenAI about AI-generated anime clips circulating online and called on the company to avoid actions that may infringe domestic copyright rules.
Officials described anime and manga as cultural assets that must be protected as generative technology evolves. OpenAI has previously said it plans to give creators more control over how their works are used and explore future revenue-sharing programs.
However, Japanese rights holders maintain that these plans do not address the use of copyrighted material during initial model training.
Lyft CEO David Risher on paying drivers more and the shift to robotaxis