AI-powered deep nudes becoming new weapon of child-on-child abuse

A new crisis is unfolding at the intersection of technology, ethics, and childhood safety: AI-generated “deep nudes.” A study titled “The phenomenon of deep nudes—a new threat to children and adults”, published in the journal AI & Society provides the first large-scale empirical evidence of how generative AI is being misused by school-aged children in the Czech Republic to produce non-consensual pornographic content.

The survey of 27,336 primary and secondary school students reveals that nearly 3% of respondents admitted to creating deep nude images, realistic depictions of nudity fabricated using AI applications. Boys were 3.5 times more likely than girls to engage in this behavior. The study reveals a rapidly evolving landscape in which deepfakes and AI-generated pornography are not only targeting celebrities but are now proliferating in school environments, affecting minors directly.

How are AI tools being used to create deep nudes in schools?

Deep nudes, a category of deepfake pornography, involve AI applications that strip clothing from photos to produce fabricated nude images of individuals, often without their consent. These tools, readily accessible online, require only a single image to generate realistic fake nudity. The applications are mostly unregulated, many of them freely available or offered with trial versions, and most do not incorporate safeguards to prevent misuse involving minors.

The Czech study collected data between September and December 2024 through voluntary participation facilitated by schoolteachers. The students, aged 10 to 18, provided anonymous responses via an online questionnaire. Of the entire sample, 758 children admitted to creating deep nude content. The overwhelming majority of these cases were boys, who comprised 76.25% of the creators.

These findings align with documented incidents in both Europe and the United States. From Spain to California, multiple cases have emerged where teenage boys downloaded classmates’ photos from social media platforms like Instagram, used AI to generate nude versions, and circulated the images via messaging apps like WhatsApp and Snapchat. In many cases, victims were blackmailed or severely emotionally impacted.

In addition to children, celebrities have also been major targets of deepfake pornography, as highlighted in the study. High-profile examples include British media personalities and American singer Taylor Swift, whose AI-generated explicit images went viral on social media in early 2024.

Who is most likely to create deep nudes, and why?

The research revealed that boys aged 13 to 17 were significantly more likely to create deep nudes, with the highest percentage occurring at age 14. By educational background, students from Secondary Vocational Schools with Apprenticeship Certificates, typically less academically oriented institutions, were most represented among offenders, with 5.76% admitting to creating such content. In contrast, only 1.49% of students from academically elite grammar schools did the same.

Despite statistically significant differences across age and school types, the study found that these differences were small in practical terms. Still, the pattern indicates that children in lower academic tracks may be more vulnerable to risky digital behavior. This echoes previous findings on online and offline sexual risk behavior, which is often higher among youth with lower educational attainment, fewer family supports, and higher levels of impulsivity or sensation-seeking.

While the study did not explore motivations, contextual interviews and case reviews suggest that most of the AI-generated deep nudes involved female classmates. The intention appears to be either for shock value, bullying, or in some cases, extortion.

What are the psychological and social consequences?

The study highlights the deep psychological toll of such incidents. Victims of AI-generated deep nudes, especially girls, face emotional trauma, anxiety, humiliation, and long-term social isolation. The public dissemination of such content often leads to cyberbullying, reputational damage, and school absenteeism. In extreme cases, it can disrupt educational progress and undermine future career opportunities.

Legally, the creation and distribution of AI-fabricated pornographic content involving minors constitutes child pornography in many jurisdictions. However, enforcement is complicated by the rapid spread of content and the limited technical understanding among educators and parents. Victims may also face secondary trauma within families and communities, further compounding the damage.

Schools are urged to respond swiftly when such incidents arise. Recommended steps include blocking access to illicit files on school systems, involving crisis response teams, providing psychological support to victims, and contacting law enforcement where necessary. Disciplinary action and clear communication with parents are also critical components of response.

The study stresses the need for preventive education, particularly around digital ethics, media literacy, and responsible technology use. Schools should incorporate lessons on AI risks into subjects like civics and computer science, while fostering a school climate that supports open communication and early detection of misconduct.

For parents, the role is equally vital. The study recommends age-appropriate conversations about online behavior, risks of digital content manipulation, and the legal ramifications of distributing intimate material, even as a joke. Parents are also encouraged to work in tandem with schools to reinforce preventive measures.

A broader threat: AI, grooming, and legal Gaps

While deep nudes represent the most visible threat, the study warns of broader risks linked to generative AI. These include cyberbullying, sextortion, revenge porn, and online grooming, all of which can now be facilitated by AI with minimal technical knowledge. What once required months of manipulation or advanced editing can now be accomplished in minutes using publicly available tools.

Language models capable of simulating personalized conversations pose an additional layer of threat. Offenders can use AI to mimic children’s linguistic styles or imitate familiar figures to gain trust, especially in grooming scenarios.

Legal frameworks remain patchy. While the European Union’s AI Act and the forthcoming Directive on Combating Violence Against Women aim to impose transparency and criminalize non-consensual sexual deepfakes, enforcement remains inconsistent across member states. The Czech Republic is in the process of introducing new legislation that would criminalize the creation and dissemination of non-consensual AI-generated pornography, including prison sentences of up to five years.