TOKYO (TR) – Tokyo Metropolitan Police have arrested a man from Hokkaido Prefecture who is suspected of using generative AI to create hundreds of thousands of obscene images of female celebrities, reports the Sankei Shimbun (Jan. 20).
Between December 2024 and May 2025, Tatsuro Chiba, 31, is suspected of posting over a dozen obscene images online, making them viewable by an unspecified number of users.
Upon his arrest on suspicion of displaying obscene electromagnetic images on Monday, Chiba, who lives in Sapporo City, Hokkaido, admitted to the allegations. “I used the money for travel expenses to chase after celebrities,” he said.
Starting around the summer of 2023, Chiba trained a generative AI tool with images of approximately 300 people, including female idols. He then created over 500,000 fake obscene images.
The suspect is believed to have made approximately 11 million yen by posting them on a members-only site where users could view them for a fee and also creating images upon request.
To create the images, Chiba used a free generative AI tool, online articles and other sources. The money he made was used for entertainment and other expenses.
“Sexually explicit deepfakes”
Such images created using generative AI are called “sexually explicit deepfakes.” Recently, this content, not only targeting celebrities but also ordinary people, has been misused.
In one case, a male office worker made a pornographic video of a female colleague that was derived from a meeting. He then shared it with colleagues. In another instance, a middle school student made pornographic images of female students from an event album. He, too, shared the images.
This month, the government requested that the operating company of X take measures following a sharp increase in the posting of sexually explicit images made using the generative AI tool Grok.
