Japanese police have charged a man with reproducing an AI-generated image without permission. This is believed to be the first lawsuit in Japan in which an image generated by AI is treated as a copyrighted work under Japanese copyright law.
According to Yomiuri Shimbun and what was discovered was dexertothis case is related to an AI-generated image created in 2024 by a man in his 20s from Chiba Prefecture, Japan, using stable diffusion. It is suspected that a 27-year-old man (also from Chiba Prefecture) then used this image on the cover of his own commercially available book without permission.
The original creator of the image told the Yomiuri Shimbun that he used more than 20,000 prompts to generate the final image. Police suspected that the author was sufficiently involved in the creation of the AI image, and referred the case to the Chiba District Public Prosecutors Office.
Japan’s copyright law defines a copyrighted work as “a creative expression of ideas or feelings that falls within the scope of literature, scholarship, art, or music.” Regarding whether images generated by AI are protected by copyright, the Agency for Cultural Affairs stipulates as follows. said AI images generated without any human instructions or with very basic instructions are not considered to be “creative expressions of thoughts or feelings,” and therefore do not meet the requirements for copyrighted works.
However, if a person uses AI as a tool to creatively express their thoughts and feelings, the output produced by AI may be considered a copyrighted work. This will be determined on a case-by-case basis. To determine whether a particular image generated by AI can be considered creative enough to be called a copyrighted work, we need to examine the process behind its creation. The main criteria are the amount of detailed prompts, refinement of the instructions through repeated production attempts, and creative choices or changes to the output.
However, this is the first lawsuit of its kind and there is no precedent. An AI and copyright legal expert at the Fukui Bar Association told the Yomiuri Shimbun that if the prompts are sufficiently detailed and specific, the output generated by the AI could be considered a copyrighted work. Essentially, the resulting final image is evaluated by how well it is deemed to match the creator’s original vision. This depends on how detailed and specific your prompts and their modifications are. “It is important to determine whether the person provided a prompt to arrive at a particular predicted outcome,” the attorney said.
On the other hand, AI-generated content that closely resembles existing copyrighted works has recently caused a huge uproar in Japan. This is especially true for video clips produced by Sora 2 that feature famous Japanese characters. of Japanese government and organizations representing major creative powerhouses like Bandai Namco, Studio Ghibli, and Square Enix. Demands OpenAI to stop unauthorized training against Japanese IP.
Photographer: Gabby Jones/Bloomberg via Getty Images.
Verity Townsend is a freelance writer based in Japan and previously served as an editor, contributor, and translator for the gaming news site Automaton West. She also writes about Japanese culture and film for various publications.
