A total of 19 manga publishing companies and associations in Japan have jointly issued this statement.
On 31 October, 2025, some of the leading manga publishing companies and anime/manga associations issued a joint statement regarding AI and copyrights infringement. The companies and associations who participated include Kodansha, Shogakukan, KADOKAWA, and many more. The main issue raised in the statement was the usage of AI to infringe on copyrighted art, and how to stop data from being stolen to train AI models.
The manga publishers joined hands to push back on the unchecked usage of AI
The report begins with the fact that in October 2025, OpenAI’s latest video generation platform Sora2 was launched. Unfortunately, a lot of videos and pictures that were made with this model and posted on social media were strongly similar to previously existing creative work. The Sora2 platform has the ability to generate, publish, and publicly transmit copyrighted work, unless the rights holder explicitly requests to opt-out.
According to the statement, the publishers think that this not only violates the principles of Japan’s copyright laws, but also the WIPO (World Intellectual Property Organization) copyright treaty. Although OpenAI’s management has hinted that Sora2 will switch to an opt-in system, the policy has not been confirmed yet. Moreover, the publishers think that with the rampant rise of different generative AI platforms, nothing is ever sure. This is why they have chosen to make their stance on the matter clear.
The statement says that the publishers welcome the advancement of generative AI and believe that, if used correctly, it can give rise to a society where more people can share the joy of creation. However, the publishers also won’t tolerate copyright infringement. Thus, they have listed some principles that may always be observed and implemented whenever AI is used to learn from human creative works. These include:
1. AI providers need to obtain necessary permission in accordance with the copyright law from the specific rights holders in order to learn, generate, and publish work based on their creations.
2. The training data should be transparent.
3. When the rights holders authorise the use of their work, they must be fairly compensated.
They believe it’s also necessary to stop situations where the AI user unknowingly generates and publishes a product based on copyrighted creations of others. The government, rights holders, AI providers, and other stakeholders need to come together and work on this issue. Providing a system which learns from creative work such as manga and anime to then recreate specific work without permission violates the Copyright Act’s principle. Thus, AI providers must also make it mandatory for AI providers to make their systems work on an ‘opt-in’ basis instead of creators having to opt-out.
The publishers also mentioned that they will take appropriate legal action against those who have infringed on their copyrighted work, whether or not they were done with AI. Instead of rejecting AI, they want to protect the efforts and dignity of the artists whose works are being stolen. The companies will continue to work closely with creators and work hard to ensure a fair, transparent, and sustainable creative environment for them in the age of AI.
The companies and associations who jointly issued this statement are: The Association of Japanese Animations, Japan Cartoonists Association, Akita Shoten, Ichijinsha Publication, Ozhora Publishing, KADOKAWA, Coamix, Kodansha, Shogakukan, Shōnen Gahōsha, Shinchosha, Square Enix, Takeshobo, TO Books, NIHONBUNGEISHA, Hakusensha, Futabasha, Houbunsha, LEED Publishing.
Source: Oricon News.