NAGOYA, (Japan): A former male teacher in Nagoya was indicted Friday (Dec 5) for allegedly possessing sexually explicit images of young girls created using generative artificial intelligence based on real images of the children.
The Nagoya District Public Prosecutors Office indicted Shota Suito, a 34-year-old former teacher at a Nagoya municipal elementary school, on suspicion of violating the Law on Punishment of Activities Relating to Child Prostitution and Child Pornography, and the Protection of Children.
This marks the first time in Japan that sexual deepfakes — obscene images created using generative AI — have been judged to constitute child pornography and the first time a person has been prosecuted under this law.
The Aichi prefectural police sent papers on Suito to prosecutors in November.
Suito is also on trial for charges including violating another law against photographing sexual postures and for allegedly sharing materials including images of minors that were secretly taken in a group chat.
According to the indictment in the latest case, Suito allegedly possessed two sexually explicit images of children at his home in March. The images had been edited and processed using a site that created AI-generated images. He had sent images of two children stored at his school to another person and had them create nude images using generative AI, according to the prefectural police.
Possession and production of sexual images of children are regulated under the child pornography law, but past judicial precedents assumed the victims were real children, which made it difficult to apply the law on deepfakes. The prefectural police determined the images constituted child pornography, citing factors such as identifying the real children from the faces in the sexual AI-generated images.
“Determining that sexually explicit images of children created using generative AI constitute child pornography challenges the premise of the child pornography law, which requires depictions of real life images,” said Associate Prof. Masaki Ueda of Kanagawa University, an expert on obscenity regulations. “This judgment is conducive to prevent cases in which parts of images of real children are used, sexually depicting them using generative AI.”
However, Ueda also said there is a risk of the lines could be blurred between creative works and actual child pornography. “It is necessary to limit the scope of the law’s application, for example, by requiring that the faces in the original image and the generated image be identical.” - The Yomiuri Shimbun
