On October 28, the Content Overseas Distribution Association (CODA), a powerful group representing many of Japan’s leading publishers and producers, formally requested OpenAI to halt the unauthorized use of its members’ copyrighted intellectual property for training the Sora 2 generative AI. This marks a significant development in Japan’s escalating controversy surrounding Sora 2.
Following Sora 2’s public launch on October 1, social media quickly filled with impressive AI-generated video clips. Many of these videos strikingly depicted copyrighted characters from popular Japanese franchises such as Pokémon, Mario, One Piece, Dragon Ball, and Demon Slayer. OpenAI CEO Sam Altman even acknowledged the strong user engagement with Japanese content in his initial blog post, noting, “We are struck by how deep the connection between users and Japanese content is!”
“The Dark Knight” pic.twitter.com/2Hk0vlJBKJ
— Justine Moore (@venturetwins) October 1, 2025
CODA, the prominent Japanese organization at the forefront of this dispute, was established in 2002 with a mission to combat piracy and champion the legal global distribution of Japanese creative works. Its impressive roster of members features titans of anime, manga, and game publishing, including Bandai Namco, Square Enix, Studio Ghibli, Cygames, Toei Animation, Kadokawa, and Aniplex (a Sony Music Entertainment Japan subsidiary).
CODA’s official statement asserts that a significant portion of Sora 2’s generated output bears a striking resemblance to existing Japanese content and images. This resemblance, they claim, is a direct consequence of Japanese copyrighted material being utilized as machine learning data without explicit authorization. Consequently, CODA argues that Sora 2’s ability to replicate specific copyrighted content through its learning process very likely constitutes copyright infringement.
In their formal communication, CODA presented two key demands to OpenAI. First, they insisted that the content belonging to their members must no longer be employed for Sora 2’s machine learning processes without obtaining prior consent. Second, they urged OpenAI to address all future copyright infringement claims and inquiries from CODA members regarding Sora 2’s generated outputs with utmost sincerity and diligence.
Previously, on September 29, it was reported that OpenAI had reached out to various studios and talent agencies just a week before Sora 2’s launch, offering them an opt-out option for their content. However, details on which specific studios were contacted within this brief window, particularly if Japanese rightsholders were included, remain vague. On the day Sora 2 launched, Japanese politician and lawyer Akihisa Shiozaki voiced significant concerns on social media about the legal implications of the AI, highlighting the urgency for Japanese companies to act. He also observed that Sora 2 appeared to struggle with generating certain American IPs, such as Disney and Marvel characters. Days later, on October 4, OpenAI CEO Sam Altman publicly stated that the company intended to ‘let rightsholders decide how to proceed’ concerning Sora 2’s output, acknowledging the possibility of ‘edge cases’ where character depictions might inadvertently appear.
While CODA’s statement recognized Sora 2’s opt-out mechanism, it firmly clarified that Japan’s copyright laws mandate prior permission for using copyrighted works. There is no legal provision that allows for avoiding infringement liability through retrospective objections. This implies that, in CODA’s view, OpenAI has already committed copyright infringement by utilizing Japanese intellectual properties for machine learning without explicit authorization from the rightsholders.
Earlier in October, the Japanese Cabinet Office itself made a formal appeal to OpenAI, urging the company to cease any infringement on copyrighted Japanese intellectual properties. This government intervention sparked widespread discussion on social media among Japanese politicians and the public, with some legislators even proposing that Japan consider implementing provisions under its AI Promotion Act if OpenAI fails to voluntarily address these concerns.
The next steps are uncertain, as the world awaits OpenAI’s official response to CODA’s latest, emphatic plea.
(Photo credit: Florian Gaertner/Photothek via Getty Images)
Verity Townsend is a Japan-based freelance writer with extensive experience covering games, Japanese culture, and cinema.