(© Yuliia - stock.adobe.com)
AI Fake Nudes Are Not Limited to Celebrities, Survey Finds
In A Nutshell
- Roughly 1 in 14 adults surveyed in Australia, the UK, and the US reported that someone used AI to create, share, or threaten to share a fake nude or sexual image of them.
- Men reported being targeted more often than women, with 10.2% of men and 3.8% of women reporting AI-generated image abuse.
- Among victims, women reported more negative reactions, including more fear for their safety and more sadness.
- Most victims blocked or reported the person responsible, but some said platforms and police did little to help.
A text from a friend asks whether the attached photo is real. The face is the recipient’s own, pasted onto a body that was never photographed, in a scene that never happened. A new survey of adults in Australia, the United Kingdom, and the United States found that roughly 1 in 14 respondents reported a similar experience: someone had used AI to create, share, or threaten to share a fake nude or sexual image of them without permission.
Researchers from RMIT University in Melbourne and Google ran the survey in July and August 2024, then published the results in the International Journal of Human-Computer Interaction. Their numbers describe an everyday digital hazard, not a problem confined to celebrities. Men reported being targeted more often than women, yet women described far more fear and sadness afterward.
Rates of AI-generated abuse were also higher among younger adults, LGBTQ+ respondents, people with disabilities, and people of color. All figures cover only abuse that victims knew about, so the real total is probably higher.
Survey of 7,230 Adults Shows AI Fake Nude Photos Reach Ordinary People
Market research firm Sago recruited 7,230 adults through online survey panels, about 2,400 in each country. Women and men were split roughly evenly, and ages ranged from 18 to over 65. Researchers adjusted the results so the sample matched each country’s population by age, gender, and race. Respondents were asked whether, since turning 18, anyone had created a fake nude or sexual image of them, shared one, or threatened to share one. Anyone who said yes was asked a follow-up: was AI, or “deepfake” technology that swaps one person’s face onto another body, involved?
Overall, 15.1% reported abuse involving digitally altered images, and 7.0% said AI was specifically involved. Fakes like these can come from “nudify” apps, which strip clothing from a real photo while keeping the person’s face. Another 8.0% reported fakes made without AI or were unsure, and those cases stayed out of the AI count.
Rates were highest in the UK at 9.3%, followed by Australia at 7.7% and the US at 6.5%. Researchers could not say why, though they pointed to possible differences in cultural norms, online habits, and willingness to disclose abuse.
Women Report More Negative Reactions to AI Fake Nude Photos
Among men, 10.2% reported AI-generated image abuse, compared with 3.8% of women. Afterward, though, women reported far rougher reactions. Among victims, about 88% of women described only negative feelings, versus 71% of men. Roughly 4 in 10 women said they feared for their safety, compared with fewer than 2 in 10 men, and about a third of women felt sad, versus about 1 in 10 men.
Some men took it lightly. About 1 in 8 found it funny, compared with fewer than 1 in 25 women. One man, targeted by a blackmail attempt using fake pictures, said, “I just laughed.”
Why would men report more abuse yet milder reactions? The authors offered a possible explanation the survey could not confirm. Images used in pranks, bullying, or blackmail are more likely to be sent straight to the target, so men may find out more often. Fakes made for private gratification may never reach the person pictured, and earlier research on deepfake porn sites found that nearly all targets were women. Abuse of women could be hidden rather than rarer.
Most Victims Knew the Person Responsible Only Online
Of those who experienced AI-generated abuse, 55.6% said the person responsible was someone they knew only online. Roughly 3 in 10 had met the person in real life. Asked to describe the relationship, both men and women most often chose “stranger,” which about 1 in 5 victims picked. The authors read that as a sign that an online-only contact can still feel like a stranger.
One earlier study of image-based abuse more broadly found that most perpetrators were known to the victim, almost half a current or former partner. A man from the UK wrote that “my ex made an explicit deepfake photo of me as revenge as he was mad that I broke up with him.” Women named a former partner somewhat more often than men did, and men more often named a friend, though the numbers were small.
Some Victims Say Platforms and Police Did Little to Help
Most victims acted. Blocking the person on a platform was the most common response (69.2%), followed by reporting the person to the platform (56.8%). In open-ended answers, people described embarrassment, doubt that a report would matter, and others dismissing a fake as “not real.” One man wrote, “I didn’t report or seek help because I thought it would be pointless.” A woman from the UK put it more bluntly: “Nobody cared, believed me or listened.”
A man who reported images to Facebook said the platform “didn’t do anything but suggest I block them.” Police responses varied: one participant obtained a protective court order, another said the offender faced charges, and several mentioned jail time.
Being fake does not make an image harmless. Ordinary adults are the ones reporting this abuse, and some who asked platforms or police for help came away with nothing. Reporting systems built around real photos need to treat AI fakes just as seriously.
Disclaimer: This article summarizes a single peer-reviewed survey and is intended for general information only. It is not medical, psychological, or legal advice, and survey findings describe groups of people, not any one person’s experience. Anyone affected by image-based abuse should consider contacting a qualified professional or a local support service.
Paper Notes
Limitations
Because the survey relied on self-reports, it captured only abuse that victims knew about; fakes created and kept privately would not appear. Respondents were asked about their entire adult lives, so some incidents may predate recent advances in AI, and some respondents may have misclassified whether AI was involved. Only adults 18 and older took part, though separate research suggests teenagers and young adults may face higher risk. Participants came from online survey panels in three high-income, English-speaking countries, which may limit how well results apply elsewhere. Perpetrator identity, gender, and relationship rested on victims’ perceptions, which can be unreliable when scammers hide behind fake online identities. The survey did not ask respondents to rate how severe or long-lasting the harms were, so harm comparisons reflect which feelings and impacts people selected. The researchers did not file their analysis plan in advance, so the analyses are exploratory, meaning the patterns need confirming, and the explanation for the gender gap is the authors’ hypothesis, not a tested finding.
Funding and Disclosures
The research was supported by an unrestricted grant from Alphabet to RMIT University, and the paper’s funding statement describes it as an unrestricted gift by Google. Three of the four authors, Rebecca Umbach, Renee Shelby, and Kwynn Gonzalez-Pons, are employed by Alphabet, Google’s parent company, and hold a financial interest in it. The authors state that all participants and the RMIT ethics review board were informed of the funding source and the authors’ employment affiliations before the study took place.
Publication Details
Title: “That Experience Still Haunts Me to This Day”: A Multi-Region Survey on AI-Generated Image-Based Sexual Abuse Victimization. Authors: Nicola Henry (Social Equity Research Centre, RMIT University, Melbourne, Australia), Rebecca Umbach (Trust & Safety, Google LLC), Renee Shelby (Google Research, Google LLC), and Kwynn Gonzalez-Pons (Trust & Safety, YouTube Entertainment Media LLC). Journal: International Journal of Human-Computer Interaction (Taylor & Francis). Published online: September 8, 2026, as an open access article under a Creative Commons Attribution 4.0 license. DOI: 10.1080/10447318.2026.2723677. Corresponding author: Nicola Henry, [email protected].
