OpenAI’s latest AI model, Sora, has created a storm. It uniquely employs copyrighted content created by Japan’s own Studio Ghibli to create new interpretations of the celebrated animation studio’s work. This practice has driven a coalition of prosecutors, public defenders, and criminal justice reform advocates to outrage and action. Japan’s Content Overseas Distribution Association (CODA) makes a strong case that these actions thus far violated Japan’s copyright laws.
Studio Ghibli, renowned for its critically acclaimed animated films, has not publicly addressed the rise of AI-generated interpretations based on its creations. The studio’s co-founder, anime auteur Hayao Miyazaki, has spoken out in the past in scorching terms against AI-generated material. In 2016, he went on the record about his extreme disdain for AI created 3D animation. His response exposes a pervasive bias about the contributions of artificial intelligence to the arts and humanities.
In fact, according to CODA, Japanese copyright law requires that you obtain permission in advance in order to use copyrighted works. This legal framework means that OpenAI’s decision to train its models on Studio Ghibli’s content without explicit consent may contravene established regulations.
“I can’t watch this stuff and find it interesting.” – Hayao Miyazaki
The issue of copyright infringement in the context of AI training is further complicated by a recent ruling from U.S. federal judge William Alsup. As a result, the court decided that Anthropic did not infringe the law when training its AI on copyrighted books. The corporation came under fire for illegally pirating the books. This decision highlights the glaring legal loophole in U.S. copyright law that permits the unfettered training of AI. The legal landscape has barely shifted since 1976.
OpenAI has been slammed from all sides. From CODA to Nintendo to the estate of Dr. Martin Luther King, Jr., creators large and small have spoken out against their copyrighted works being used without permission. The organization has not answered questions regarding whether it intends to heed CODA’s call to stop its in-progress campaigns.
Furthermore, OpenAI’s image generator has been utilized to transform users’ selfies and pet photos into styles reminiscent of Studio Ghibli’s iconic animation. OpenAI CEO Sam Altman has certainly jumped on the trend. He even changed his profile picture on X to a “Ghiblified” version of himself!
“I feel strongly that this is an insult to life itself.” – Hayao Miyazaki
This ongoing debate over the legitimacy of AI-generated interpretations of protected works has been at the heart of a deepening rift with the creative community. With no clear precedent to guide judicial interpretation of copyright law in this digital age, many creators and companies have begun advocating for more stringent regulations. Japan’s copyright system provides no mechanisms for companies to escape liability for infringement based on subsequent objections. This framing puts even more burdens on companies like OpenAI.