Studio Ghibli and Publishers Demand OpenAI Halt Training on Their Work

Copyright Concerns Raised by Japanese Publishers

A coalition of Japanese publishers, including representatives from Studio Ghibli, recently addressed a letter to OpenAI. They are requesting the company cease utilizing their copyrighted content for the training of its artificial intelligence models without explicit authorization.

Impact on Studio Ghibli

Studio Ghibli, renowned for animated features such as “Spirited Away” and “My Neighbor Totoro,” has experienced notable effects from OpenAI’s generative AI technologies. Following the March release of ChatGPT’s image generation capabilities, a trend emerged where users requested recreations of personal photos in the distinctive style of the studio’s films.

Even OpenAI’s CEO, Sam Altman, participated in this trend by modifying his profile picture on X to an image rendered in the Ghibli aesthetic.

CODA's Request Regarding Sora

With increased access to OpenAI’s Sora app and its video generation features, the Japan’s Content Overseas Distribution Association (CODA) has formally asked OpenAI to refrain from employing its members’ content for machine learning purposes without prior consent.

OpenAI’s Approach to Copyright

This appeal stems from OpenAI’s current strategy of seeking forgiveness rather than permission regarding copyrighted material. This approach has facilitated the creation of images and videos featuring protected characters and likenesses of individuals who are no longer living.

This practice has already prompted complaints from organizations like Nintendo, and from the estate of Dr. Martin Luther King, Jr., who could potentially be misrepresented through deepfakes generated by the Sora app.

Legal Considerations

The decision to cooperate with these requests rests with OpenAI. Failure to do so could result in legal action, although the legal landscape in the United States regarding the use of copyrighted material for AI training remains ambiguous.

Current copyright law, which has not been significantly updated since 1976, provides limited guidance for judges interpreting these cases.

However, a recent ruling indicated that Anthropic did not violate the law by training its AI on copyrighted books. The company was, however, penalized for the initial unauthorized acquisition of those books.

Japanese Copyright Law

CODA asserts that the situation may be viewed differently under Japanese law.

“In instances, such as with Sora 2, where specific copyrighted works are replicated or similarly generated as outputs, CODA believes that the act of reproduction during the machine learning process may constitute copyright infringement,” the organization stated.

“Japan’s copyright system generally requires prior permission for the use of copyrighted works, and does not provide a mechanism to avoid liability through subsequent objections.”

Hayao Miyazaki’s Perspective

Hayao Miyazaki, a pivotal figure at Studio Ghibli, has not publicly commented on the proliferation of AI-generated interpretations of his work.

However, in 2016, upon viewing AI-generated 3D animation, he expressed strong disapproval, stating, “I can’t watch this stuff and find it interesting. I feel strongly that this is an insult to life itself.”

He described the experience as “utterly disgusted.”