Researchers report that more than half of the U.S. teens surveyed used artificial intelligence tools to create fake sexualized nude images.
This study reframed these images as a common part of teen digital behavior, rather than rare or marginal misuse.
Many teens create fake images
Within the anonymous group of 557 teenagers, this activity appeared to be widely integrated into everyday online interactions, rather than being limited to isolated cases.
After analyzing these responses, Chad Steele of George Mason University (GMU) determined that 55.3 percent created at least one such image.
Additionally, more than half of teens reported receiving these images, demonstrating how creation has quickly spread to everyday sharing among peers.
The pattern points to behaviors that are already being normalized at scale and sets the stage for deeper questions about consent and harm.
Unlike image generators that build scenes from prompts, or software that makes a clothed photo appear nude, AI starts with a photo of a real person.
This design reduces the effort required to create sexualized images, as a single existing selfie becomes the material in seconds.
Steel found that teens use these apps more than extensive image-making systems. This was important because the targets were usually recognizable.
Once the real face establishes the image, it becomes difficult to dismiss embarrassment, coercion, and rumor-spreading as fantasy.
Teens facing non-consensual harm
Harms appeared about as often as participation, and the difference between the two was smaller than many adults expected.
In the survey, 36.3% said someone had created a sexualized AI image of them without their permission.
A further 33.2% said someone shared one of those images, turning a private violation into a social event.
Creation and distribution are important as separate wounds. This is because stopping one does not automatically stop the other.
Behaviors common to all teens
Across race, age, and most other categories, the behavior appeared to be widely dispersed rather than clustered into one clear subgroup.
Male participants still reported higher rates in several categories, including creating or sharing images of themselves, co-workers, and adults.
Female participants approximated the overall pattern, overturning the idea that only boys participate.
Programs that target a single stereotype miss many of the behaviors that are actually captured in research.
Teenagers also participate
Age did not provide clear protection within the study. Results showed that 13- and 17-year-olds reported similar levels of use and victimization.
Schools could read this finding as an example of early lessons on consent, privacy, and image sharing, before middle school habits become ingrained.
Elsewhere, the UK’s communications regulator released a report in 2025 that found half of children aged 8 to 17 had used an AI tool.
Familiarity with AI no longer begins in late adolescence. This helps explain why waiting until high school can cause you to miss out on that moment.
AI turns this into the new sexting
Although previous teen sexting research did not involve AI, it helps show how far the behavior has changed.
A 2018 study of 39 studies found that 14.8 percent of adolescent sexting was sending and 27.4 percent receiving.
Images can be generated from prompts or edited from real photos, reducing effort and blurring responsibilities.
“Teens are no longer just digital natives, they are AI natives. ‘Nudity’ and GenAI apps are their new ‘sexting,’ only with more difficult questions around consent,” Steele said.
Teenagers’ actions are against the law
Even if the image is completely synthetic, federal law is already broader than many teenagers probably realize.
Under the law, lewd sexual images involving minors can be illegal even if the biological child is not present.
This creates a conflict with peer-to-peer behavior, as some teens may treat these interactions as flirtation or experimentation.
Schools, parents and lawmakers still face the more difficult question of how to stop the harm without pretending that such behavior is uncommon.
part of a larger problem
Beyond research, sexual images created by young people already make up a large portion of online abuse reports.
According to the Internet Watch Foundation’s 2023 report, 92% of the child sexual abuse images it identified and helped remove were self-produced.
AI has not created an urge to create or exchange intimate images, but it has made them easier to edit, forge, and redistribute.
Prevention cannot focus solely on AI when social behavior is clearly older than the software itself.
Research limitations
Although the study, taken in January 2025, provides solid numbers on hidden behavior, it still leaves important blind spots.
GMU conducted the study only with English-speaking U.S. teens ages 13 to 17 and required parental consent before participation.
Adult perpetrators, younger children, and many smaller subgroups were outside the plan, so some harms may be greater or different.
Even with these limitations, the results are too large to treat as a narrow problem awaiting additional data.
This is not a fringe behavior confined to a few reckless teens, but a widespread digital practice shaped by issues of consent, social pressure, and harm.
This reality makes early education, stronger safeguards, and better support for victims not only important, but outdated.
The research will be published in a journal pro swan.
—–
Like what you read? Subscribe to our newsletter for fascinating articles, exclusive content and the latest updates.
Check out EarthSnap, a free app from Eric Ralls and Earth.com.
—–
