Studio Ghibli and Japanese publishers urge OpenAI to stop training on their creative work

75

In a growing pushback against unlicensed AI training, Studio Ghibli and a coalition of Japanese publishers have urged OpenAI to stop using their creative works — including films, books, and illustrations — to train generative AI models.

The group, which includes prominent publishing houses such as Kadokawa, Shogakukan, and Shueisha, has formally requested OpenAI to clarify how copyrighted Japanese content is used in datasets that underpin systems like ChatGPT and DALL·E.

Citing concerns over copyright infringement and cultural integrity, the creators argue that AI models built on unlicensed data threaten the livelihood of artists and undermine Japan’s long-standing commitment to respect for intellectual property. Studio Ghibli, renowned for classics like Spirited Away and My Neighbor Totoro, said that its works “should not be used without explicit consent or context.”

The statement follows similar moves in the U.S. and Europe, where authors and creative guilds have sued AI developers over data scraping and content replication. Japan’s appeal stands out for framing the issue around cultural preservation as much as legal protection.

OpenAI has said it continues to review its data sourcing practices and maintains that its models comply with fair use standards. However, experts predict that Japan’s unified stance could pressure global AI firms to rethink licensing and transparency frameworks for creative content.