CODA Demands OpenAI Cease Use of Japanese Content for Sora 2 Training
The Content Overseas Distribution Association (CODA), a prominent Japanese content trade group that includes renowned animation studio Studio Ghibli among its members, has formally requested that OpenAI halt the use of its content to train the AI model Sora 2. This demand comes in light of concerns that the model may infringe on Japan’s copyright laws by utilizing Japanese intellectual property without proper authorization.
CODA’s mission is to promote the global expansion of Japanese content while implementing anti-piracy measures. Last week, at the request of its member companies, CODA submitted a written appeal to OpenAI regarding Sora 2, which was launched in September. In their letter dated October 27, CODA stated, “We have confirmed that a significant portion of Sora 2’s output closely resembles Japanese content or imagery.” They believe that this similarity arises from the use of Japanese works as training data.
The group further explained that when Sora 2 reproduces or generates outputs resembling specific copyrighted works, it may constitute copyright infringement. This concern is heightened by OpenAI’s opt-out system, which requires copyright holders to notify the company if they do not want their works included in the generative AI model. This approach has drawn criticism from U.S.-based entertainment organizations, including the Motion Picture Association (MPA).
OpenAI CEO Sam Altman has suggested that Sora 2 will offer rights holders “more granular control over the generation of characters,” paralleling the opt-in model for likenesses but with additional safeguards. However, CODA argues that under Japan’s copyright framework, prior permission is generally mandatory for the use of copyrighted materials. They assert that there is no provision allowing individuals to evade liability for infringement through subsequent objections.
In their letter, CODA firmly requested that OpenAI cease utilizing content from its members for machine learning without prior permission. They also urged OpenAI to “respond sincerely to claims and inquiries from CODA member companies regarding copyright infringement related to Sora 2’s outputs.” As of now, OpenAI has not publicly responded to these concerns.
Earlier this year, following the launch of OpenAI’s latest image generator integrated into GPT-4o, users began creating numerous images inspired by Studio Ghibli’s distinctive style. Altman humorously remarked on the trend, reflecting on the rapid shift from years of hard work to suddenly receiving a flood of messages showcasing AI-generated Ghibli-style art.
Notably, Hayao Miyazaki, co-founder of Studio Ghibli, has previously expressed his strong opposition to AI-generated animation. During a 2016 demonstration of AI animation technology, Miyazaki stated, “I am utterly disgusted,” emphasizing his desire to distance his work from such technology and labeling it as an insult to the essence of life.
Founded in 2002, CODA represents a diverse array of content creators across various industries, including anime, film, television, music, video games, and publishing. Its membership includes notable companies such as Sony Music Entertainment (Japan) Inc.-owned Aniplex, NTT DoCoMo Studio & Live, Kadokawa, Shogakukan-Shueisha Productions Co., Square Enix, Studio Ghibli, and many others.
The letter from CODA to OpenAI, dated October 28, has been reported by various tech news outlets, including The Verge.
Stay tuned for updates as this situation develops, and learn more about the implications of AI on creative industries.
