Japanese publishers demand OpenAI stop using copyrighted anime content for AI training

A leading Japanese trade group representing major publishers, including Studio Ghibli, has called on OpenAI to halt the use of Japanese copyrighted material in training its artificial intelligence systems.

The Content Overseas Distribution Association (CODA) sent a formal letter to OpenAI chief executive Sam Altman, urging the company to stop using Japanese animated works for its Sora 2 AI video generation platform.

According to the letter, CODA found that a “large portion” of Sora 2’s generated content “closely resembles Japanese content or images,” suggesting that copyrighted Japanese works were used during training. The organisation said it believes this could amount to copyright infringement.

“In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement,” the letter stated.

CODA also criticised OpenAI’s opt-out system for copyright holders, arguing that it conflicts with Japanese law. The association said that under Japan’s copyright system, creators’ prior consent is required before their work can be used. “There is no system allowing one to avoid liability for infringement through subsequent objections,” the statement added.

OpenAI has yet to issue a response to CODA’s request.

Earlier this year, OpenAI integrated its most advanced image generator into ChatGPT-4o, enabling users to create images in the style of Studio Ghibli’s iconic films, including Spirited Away, Howl’s Moving Castle, My Neighbour Totoro, Kiki’s Delivery Service, and The Wind Rises.

The renewed debate over AI’s use of copyrighted material has reignited public interest in Studio Ghibli co-founder Hayao Miyazaki’s past criticism of AI-generated art. A resurfaced interview clip shows the filmmaker reacting strongly to an AI animation demonstration, saying he was “utterly disgusted” and describing the technology as “an insult to life itself.”

CODA’s intervention adds to the growing global tension between creators and AI developers over how intellectual property is used in training powerful generative models.