Studio Ghibli, Japan’s top studios demand OpenAI to stop training Sora 2 on its art

Studio Ghibli and other major Japanese entertainment companies are demanding OpenAI stop training Sora 2 on their content, a fundamental clash between Silicon Valley’s data-scraping practices and Japan’s strict copyright protections.

The Content Overseas Distribution Association (CODA), representing powerhouses like Bandai Namco, Square Enix, and Studio Ghibli, submitted a formal request to OpenAI on October 27, 2025. The organization confirmed that Sora 2’s outputs closely resemble specific Japanese copyrighted works.

“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,” CODA stated in its letter.

OpenAI operates on an opt-out system where copyright holders must actively request exclusion. However, Japanese law requires prior permission for using copyrighted material.

“Under Japan’s copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections,” CODA explained.

This confrontation has been building since March 2025, when ChatGPT users flooded social media with “Ghibli-style” AI images. OpenAI CEO Sam Altman even changed his profile picture to a Studio Ghibli-inspired portrait.

The discourse is particularly sensitive for Studio Ghibli. Co-founder Hayao Miyazaki has been vocal about his opposition to artificial intelligence (AI) animation since 2016.

“I am utterly disgusted,” Miyazaki said when shown an AI-generated animation. “I feel strongly that this is an insult to life itself.”

Sora 2 launched on September 30, 2025, quickly generating viral videos featuring characters from Dragon Ball, Mario, and other Japanese franchises. The rapid proliferation of this content prompted Japan’s government to formally ask OpenAI to stop replicating Japanese artwork.

CODA’s request goes beyond asking OpenAI to stop unauthorized training. The organization also demands that OpenAI “responds sincerely to claims and inquiries from CODA member companies regarding copyright infringement related to Sora 2’s outputs.”

OpenAI has not responded to requests for comment about the CODA letter.

The confrontation could set important precedents for AI development globally. While U.S. copyright law hasn’t been updated since 1976, leaving AI training in legal gray areas, Japan’s clearer requirements could force changes in how AI companies gather training data.

OpenAI is already facing lawsuits from The New York Times and other publishers. The Japanese challenge represents a new front in the battle over AI training practices.