Studio Ghibli, Square Enix and More Tells OpenAI To Stop Using Its Art to Train Sora 2
Japanese rights holders push consent-first training as anime and game giants unite against opt-out practices.
Summary
- Japan’s CODA formally requested OpenAI to cease using Japanese IP to train Sora 2, citing concerns over copyright infringement
- The demand, a collective push from rights holders, follows viral Sora 2 outputs that mimic protected works and styles, like those of Studio Ghibli
- Signatories include Studio Ghibli, Square Enix, Bandai Namco, and Shueisha, arguing that Japan’s law requires prior permission for using copyrighted content
Japan’s Content Overseas Distribution Association (CODA) delivered a written request to OpenAI on Oct. 27, asking for changes to how Sora 2 operates. It is the clearest collective pushback yet from major Japanese rights holders.
The group says Sora’s outputs mirror protected works and that using copyrighted material for training may infringe. CODA adds, “prior permission is generally required for the use of copyrighted works”. CODA asks OpenAI to stop using member content for machine learning without consent and to respond to member claims tied to Sora outputs.
Signatories span anime, manga, and games, including Studio Ghibli, Square Enix, Bandai Namco, Aniplex, Kadokawa, and Shueisha. The stakes are cultural and commercial.
Sora 2 launched September 30 with new generation controls and instant virality. Japan’s permission-first framework clashes with opt-out policies that leave IP owners chasing violations after the fact.
For culture-watchers, the fight is about style ownership and consent. Ghibli-style trends turned a meme into a creative red line for Japan’s most influential studios.
