In Japan they seem to be tired of images generated with artificial intelligence that resemble, perhaps too much, the mythical works of Japanese origin. We are referring, of course, to images and videos created with AI that seek to reimagine any photo, person or character in “Ghibli style” or similar. An anti-piracy organization in Japan has demanded that OpenAI cease what they claim is a copyright violation.
Japan studies against AI. CODA is a Japanese anti-piracy organization that includes companies such as Studio Ghibli, Toei Animation, Bandai, Toho and Square Enix among its members. The organization has published a letter demanding OpenAI stop using its members’ original content to train Sora 2, the OpenAI tool responsible for generating realistic videos with artificial intelligence.
Some of Studio Ghibli’s most legendary films. (Images: Studio Ghibli)
In its letter, CODA (whose acronym stands for the Overseas Content Distribution Association) claims to have confirmed that “a large portion of the content produced by Sora closely resembles Japanese content or images.” This, according to the organization, would be the result of having used copyrighted content to train artificial intelligence.
Amazon for $38 billion ” width=”375″ height=”142″ src=”https://i.blogs.es/2145e5/openai-data-center/375_142.jpeg”/>
What Japanese studies ask for. CODA’s demands are clear: that OpenAI not use its members’ content to train its artificial intelligence model. And also, that OpenAI respond to the demands and complaints of the companies that are part of the Japanese organization about the Sora 2 videos.
The government also pressures. In mid-October, the Japanese government had already spoken out against the use of copyright-protected content by OpenAI to train its artificial intelligence. Minoru Kiuchi, Japanese minister responsible for intellectual property strategy in the country, asked OpenAI not to violate the copyrights of Japanese intellectual properties. According to Minister Kiuchi, manga and anime are “irreplaceable treasures” that Japan offers the world.
2025, the year of “Ghibli-style” images. Last March OpenAI enabled GPT-4o-based image generation, and quickly “Ghibli-style” or “anime-style” images became extremely popular. However, the claims from CODA and its members, in addition to the Japanese government’s request, are especially directed at Sora 2 and its video generation capabilities.
Although the results are far from perfect, social networks have been filled with these types of unofficial videos made with AI, which for companies such as Bandai Namco, NHK, Wowow, Aniplex and many others represents a violation of their copyright. At the time of publishing this article, OpenAI has not yet responded to the Japanese studios’ request.
Cover image | OpenAI / Image created with artificial intelligence
In WorldOfSoftware | The “AI slop” turned into art. A Chinese creator is copying the absurd aesthetics of generative AI, and it’s hilarious
In WorldOfSoftware | OpenAI knows that ChatGPT is causing serious mental health problems for some users. And he is already “correcting” it
