TOKYO - As generative AI becomes more widespread, reports of sexual deepfake abuse are increasing around the world. Behind the surge are websites that use artificial intelligence to transform ordinary photos into fake nude images and videos—often without the subject’s consent.
A Japanese news program reached out to several such websites. One responded via email, providing insight into its policies and attitude toward the issue.
Although the platform had no Japanese-speaking staff, the response was sent in English. The operator acknowledged that sexual deepfakes are a global concern and thanked the program for addressing the topic.
According to the site’s user policy, uploading images of individuals under 18 is strictly prohibited. When asked how such rules are enforced, the operator stated that the platform treats the protection of minors seriously. If a user uploads an image of a minor, the system immediately blocks all further content generation from that user. This applies to both image and video generation functions currently being rolled out.
While claiming to have protective measures in place, the site also promoted its AI technology, calling it a major step in the evolution of generative AI. The operator described the platform as one of the first to implement these capabilities, with growing interest from users in Japan and abroad.
In 2023, a scandal erupted in Spain when fake nude images of at least 20 schoolgirls were created and shared on social media. The images were reportedly generated using this same platform’s services.
At the time, the site listed its headquarters in the British Virgin Islands. By 2024, that had changed to Buenos Aires, Argentina. However, a report by German magazine Der Spiegel found that internal documents were written in Russian and the company’s email infrastructure was also based in Russia. The reply sent to the Japanese media outlet contained non-English elements in the auto-generated email header, further supporting these findings.
Japan currently has no law specifically banning the creation of deepfakes. However, legal experts note that existing laws—such as those prohibiting child pornography, copyright infringement, and unauthorized use of personal images—may apply in some cases.
If deepfake content is distributed within Japan, Japanese law could be enforced, legal analysts said. Although Japan’s adult entertainment law allows certain adult websites to operate under specific conditions, those that fall outside those boundaries face serious legal risks. Pursuing legal action against foreign operators, however, would require diplomatic assistance and cooperation from law enforcement in other countries.
Legal scholar Nishida said that Japan’s current regulations are weak. In the case of real minors, sexual deepfakes may be treated as child pornography. However, fictional depictions of minors do not fall under existing laws, creating a loophole that is at odds with public expectations.
Some countries, including the United States, have passed laws that ban non-consensual sexual deepfakes altogether. Despite early concerns about overreach, the legislation gained strong bipartisan support and was ultimately signed into law.
The U.S. follows a notice-and-takedown system, requiring platforms to remove flagged content. Yet as long as the image involves a fictional subject, there may be no legal victim, making regulation difficult.
Major platforms already employ automatic content detection and human moderators to block explicit material. Whether regulation should take the form of guidelines (soft law) or enforceable legislation (hard law) remains a topic of debate.
In Japan, the AI Strategy Council recently released an interim report on managing AI risks. While acknowledging the dangers, the council cautioned that heavy-handed regulation could stifle innovation and weaken global competitiveness. It recommended limiting regulation to cases where voluntary compliance by businesses cannot be expected.
Japan’s approach to AI remains largely promotion-focused. As a result, services operating abroad but accessed by Japanese users fall into a legal grey zone. If servers are located overseas, enforcement becomes even more difficult.
While technology alone cannot fully solve the issue, experts say it must play a central role in developing solutions. The challenge is finding the right balance between innovation and the protection of individual rights.
Source: ABEMA
