Common Sense Media finds 44% of US teens believe they have seen AI-generated sexual content
Common Sense Media and Burson found that 44% of US teens surveyed believed they had seen AI-generated sexual content
AI-generated nudes and sexual images are no longer an abstract online risk for many US teenagers. New research from Common Sense Media and Burson finds the material appearing in social feeds, private messages and school conversations, with 24% of exposed respondents saying they had seen content depicting themselves or someone they knew.
Across the full sample of 1,314 teens ages 13 to 17, 44% said they had seen sexual content they believed was AI-generated. Among the 586 respondents who reported exposure, 79% said they had encountered it accidentally.
Nearly half of US teens surveyed said they had seen AI-generated sexual content. The researchers relied on the teenagers’ own identification of the material, so the figure does not confirm that every image or video was created using AI.
The routes of exposure were less ambiguous. Among exposed teens, 82% had encountered the content on social media, 41% on pornography websites, 23% through one-to-one text messages and 20% through group messaging apps. TikTok was named by 33%, Instagram by 27% and X, formerly Twitter, by 25%.
For 24% of exposed respondents, the subject was themselves or somebody they knew personally. That included 11% who said they had seen AI-generated sexual content depicting themselves.
The findings extended beyond what teenagers had seen online. Eighteen percent said they had created AI-generated sexual content or knew someone who had, while 8% reported making it themselves. Boys were more than twice as likely as girls to have done so, at 11% compared with 4%.
Among the 231 teens who had created the material or knew somebody who had, 63% said it depicted themselves or people they knew, including peers, friends and current or former romantic partners.
One 17-year-old boy described how little effort was required: “With no hurdles or anything you have to jump over, you can just do it right there.”
Sharing was more widespread. One-quarter of respondents had sent or shared AI-generated sexual content or knew somebody who had, while 9% said they had done so personally. Among teens who had shared it or knew somebody who had, 37% said the content had been passed around during in-person conversations at school.
Schools sit outside the first line of support
Only 20% of all respondents had discussed AI-generated sexual content with a trusted adult. More than one-third, 34%, wanted to have that conversation but did not know how to begin.
Understanding and fear of blame were central. Fifty-nine percent said knowing that an adult would understand would make them more likely to disclose an incident, while 53% wanted reassurance that they would not be blamed.
Among teens who had seen disturbing AI-generated sexual content, 48% would feel comfortable talking to close friends and 42% would approach a parent or guardian. Just 13% selected a teacher or school counselor. Nineteen percent said they would handle it alone, while 10% would feel comfortable discussing it with an AI chatbot.
The gap is not confined to help-seeking. Forty-four percent of respondents said schools or teachers had taught them nothing about identifying fake or AI-generated material online. Twenty-eight percent said the same of parents and family members.
A 17-year-old girl involved in the focus-group phase described the perceived generational divide:
“We grew as AI grew. We grew up together. Adults don't have the same life experience that we do. They don't understand how advanced and how intense AI is.”
Concern is already changing how teens present themselves online. Two-thirds of respondents, 67%, were at least somewhat worried about being depicted in a deepfake without consent. Girls reported greater concern than boys, at 71% compared with 62%.
Among those who were worried, 38% had become more careful about what they posted, 30% had made social media accounts private, 23% were posting fewer pictures and videos of themselves and 9% had deleted accounts.
Attitudes to harm were less settled. Among exposed teens, 82% agreed that creating AI-generated sexual content of somebody without permission should be illegal. However, 23% believed AI-generated nudes and sexual content were less harmful than real pornography because “no one gets hurt.”
The report notes that AI-generated sexual content depicting somebody under 18 is considered child sexual abuse material and a felony-level offense.
Respondents also believed the material was affecting how their peers understood bodies and relationships. Across the full sample, 66% thought it influenced ideas about how bodies should look, while 64% said it affected expectations of romantic partners and views of sex and consent. These figures measure teenagers’ perceptions of influence, rather than demonstrated effects.
What the study can establish
Common Sense Media describes the survey as nationally representative. Burson recruited participants through multiple opt-in online panels and weighted the results by age, gender, race and ethnicity, urbanicity and geography using US Census data.
The 12-minute online survey was conducted from November 21 to December 9, 2025. Its margin of sampling error was plus or minus 2.7 percentage points, with higher margins for demographic subgroups.
It formed part of a three-stage study conducted between September 22, 2025, and February 26, 2026. Twenty in-depth interviews informed the questionnaire, while two focus groups involving five girls and five boys were used to add context to the quantitative results. Parents or guardians provided consent and teens provided assent, with WCG Clinical overseeing institutional review board approval.
The findings therefore capture self-reported exposure, behavior and attitudes. They do not independently verify whether individual pieces of content were AI-generated or establish that exposure caused the effects respondents perceived.
Teens in the AI Era: Pornography and Sexual Content was written by Supreet Mann, Laura Zimmermann, Jenny Radesky and Michael Robb, and funded by Oak Foundation.
The report recommends that schools address deepfakes and digital consent before an incident occurs, establish response protocols and distinguish between students who are targets, bystanders, creators or sharers. It also calls for schools to ensure that approved AI tools cannot generate sexual or intimate images and to block unauthorized AI apps and websites on school-issued devices.
