Nearly 20% of teens have created AI-generated sexual content themselves, or know someone who has. Almost as many say they’d rather talk to a chatbot about it than a school counselor.
That second number is the real story.
The gap nobody built a policy for
Common Sense Media’s latest survey covers more than 1,300 teens aged 13 to 17, and it tracks how three years of generative AI have reshaped an old problem. Seventy-five percent of teens have seen online pornography, a number that’s barely moved since the organization’s 2023 report. What’s changed is the delivery mechanism.
Nearly half of respondents say they’ve encountered sexual content they believe was AI-generated: deepfakes, fake nudes, output from “nudification” apps that strip clothing from a photo with one upload. Lead researcher Dr. Supreet Mann puts the shift in blunt terms. These tools, she says, are “removing the agency and stripping people of their ability to control their image.” More than half of the teens who created this kind of content say it depicted someone they actually knew. A classmate, not a celebrity.
Several school districts are already dealing with legal fallout from students generating explicit fakes of their peers, and administrators are stuck in a jurisdictional no-man’s-land. The app runs on a personal phone, off campus, after hours. The two students involved still sit in the same homeroom the next morning. That proximity is exactly what makes this a campus problem whether or not the conduct happened on campus property — a tension our earlier look at teen chatbot dangers has been tracking since long before this survey landed.
Why the chatbot beats the counselor
The most uncomfortable data point isn’t about exposure. It’s about trust. Teens report feeling nearly as comfortable discussing this with an AI chatbot as with a teacher or counselor. Not because the chatbot understands the stakes better. Because it doesn’t flinch, doesn’t escalate, doesn’t call a parent.
For a teenager weighing whether to report a fake image of a friend, that low-friction, judgment-free interface can look like the safer option — even when it’s the one least equipped to actually intervene. It’s the same dynamic we flagged in how teens are actually using AI chatbots: the tool that’s easiest to talk to isn’t always the one that can help.
Nudification tools didn’t invent teen exposure to sexual content. They removed the friction that used to slow it down. A 13-year-old no longer needs to go looking for porn. The tools now generate a version of a specific classmate, on demand, with no technical skill required — which is a large part of why spotting a deepfake has quietly become a skill teenagers need before they need algebra.
| Metric | Finding | Impact |
|---|---|---|
| Peer deepfake creation | ~20% created or know someone who created AI explicit content | Shifts the issue from commercial porn to targeted peer harassment |
| Target identity | More than half depicted someone the teen personally knew | Off-campus conduct now bleeds directly into campus climate |
| Counselor vs. chatbot trust | Comfort levels nearly matched between the two | Signals fear of escalation, not lack of concern |
What’s actually being done about it
Regulation is starting to catch up, unevenly. California’s SB 243, signed in October 2025, now requires companion chatbot operators to filter out sexually explicit content reaching minors — a narrower fix aimed at a different product category, but a sign state lawmakers are willing to legislate this space at all.
For families dealing with an image that’s already circulating, the NCMEC’s Take It Down tool lets a minor or parent submit a hash of the image to help platforms detect and remove it, without uploading the image itself anywhere public. Schools updating conduct codes to explicitly cover off-campus AI generation involving classmates are ahead of where most districts sit today.
Platforms building these tools have mostly stayed out of the conversation about what happens after the image gets made. That’s starting to look less like an oversight and more like the actual policy.
Related: The Real Product AI Companion Apps Are Selling Isn’t Conversation
