On Monday, Tokyo police announced the arrest of a 32-year-old man from Himeji, Hyogo for using generative AI to create nude images of female track and field athletes — some of them as young as 12 — and posting them to social media groups. “It made me happy that I could make people happy,” the suspect reportedly told police.
Investigators say he created around 300 images and posted roughly a third of them, some at the request of a 17-year-old member of the same group, who has also been referred to prosecutors. One victim is now in her 40s; the source photo used to create her deepfake was taken when she was in her first year of middle school.
What makes the case worth attention isn’t the suspect — it’s the law he’s being charged under. Or rather, laws, plural, since Japan has none written for this specifically.
Image: Victoria Bennett | Sources: Shutterstock
Old Laws for a New Technology
Split the victims by age and the prosecution splits with them. For the adult woman in her 40s, prosecutors are pursuing defamation, a law with no particular relationship to sexual imagery, deepfaked or otherwise; it’s simply the tool available for reputational harm.
For the minors, the case falls under Japan’s child pornography law, which predates generative AI by decades but has increasingly been stretched to cover it. In June, a Nagoya court handed down Japan’s first sentence for possessing AI-generated child sexual images, applying the same law to a former elementary school teacher.
Neither law criminalizes creating a sexually explicit deepfake itself. Instead, prosecutors rely on existing offenses based on who the victim is and how the image was used. Japan’s Ministry of Justice has confirmed that AI-generated images of real children can qualify as child pornography, but outside of that context, prosecutors are left fitting deepfake cases into older laws.
The only real movement has been local. Tottori Prefecture now fines residents who create or distribute AI-manipulated images of real minors, under an ordinance that took effect this July. But there is still no national law treating deepfake sexual imagery as a crime in its own right, the way some of South Korea’s newer offenses do.
Image: Victoria Bennett | Sources: Shutterstock
Why It matters
The gap creates practical problems.
Defamation charges generally require victims to file complaints, but manipulated images often spread long before victims even know they exist. Prosecutors must also prove that an AI-generated image legally depicts an identifiable real person — a hurdle separate from proving who created or shared it.
Japan has seen a string of deepfake-related arrests over the past year, including a man accused of selling more than 500,000 AI-generated sexual images of celebrities and another accused of creating deepfakes of more than 260 female celebrities. Each case has relied on a different legal theory, exposing the same underlying reality: while generative AI has transformed how these images are made, Japan’s legal response is still built from laws written for something else.
Related Posts
- The World’s First AI Mayor is Here: Introducing Yokosuka’s Virtual Katsuaki Kamiji
- Tokyo Man Arrested After Stealing Toilet Valves To Fund Gambling Habit
Updated On August 4, 2026
