AI image tools have made synthetic media easier to create. WIRED’s review of one deepfake nude generator shows the darker side of that access: a site where visitors could see feeds of apparent uploads meant to turn real people’s photos into fake nude images.
The site is still online, and WIRED did not identify it in order to protect the women and girls shown in its feeds. What the report describes is not an abstract misuse case. It is a visible record of people being targeted without consent.
What WIRED Found
The website reviewed by WIRED included two feeds, labeled "Home" and "Explore." The feeds appeared to show photos uploaded by users who wanted to “nudify” the subjects.
WIRED saw images of girls who were clearly children. Other photos showed adults, with captions suggesting they were female friends, female strangers, classmates or romantic partners. One caption read “My gf,” attached to a mirror selfie of a young woman.
The report also described images that appeared to come from everyday social media posts. Some showed influencers popular on TikTok, Instagram and other platforms. Others appeared to be Instagram screenshots of ordinary moments, including a young woman smiling with a dessert topped with a celebratory candle.
Several photos appeared to show people who were not posing for the camera at all. One image taken from behind showed a woman or girl standing near what appeared to be a tourist attraction. Some images were cropped to focus only on women’s and girls’ chests or crotches.
Children, Classrooms And Everyday Photos
The most disturbing part of the report is the presence of minors. WIRED said several images clearly showed girls under the age of 18.
One photo showed a young girl with a flower in her hair standing against a tree. Another appeared to show a girl in a middle or high school classroom, in a photo seemingly taken discreetly by a classmate. That image was captioned “PORN.”
Another image showed young teens who appeared to be in middle school: a boy taking a selfie in what looked like a school gymnasium with two girls smiling and posing. The boy’s face was obscured by a Snapchat lens that enlarged his eyes.
WIRED notified the National Center for Missing & Exploited Children, which helps report cases of child exploitation to law enforcement, about the site’s existence.
How The Site Appeared To Operate
The site’s homepage did not show fake nude images to visitors who were not logged in. People who wanted to create and save deepfake nude images were asked to log in using a cryptocurrency wallet.
Pricing was not listed when WIRED reviewed the site. But a 2022 video posted by an affiliated YouTube page showed the website letting users buy credits to create deepfake nude images, starting at 5 credits for $5.
WIRED learned about the site through a post on a subreddit about NFT marketplace OpenSea, which linked to the YouTube page. After WIRED contacted YouTube, the platform said it terminated the channel. Reddit told WIRED that the user had been banned.
The site’s IP address, which went live in February 2022, belongs to Cloudflare. Company spokesperson Jackie Dutton pointed to the distinction between providing a site’s IP address, as Cloudflare does, and hosting its contents, which it does not.
Users were asked to log in with Coinbase, Metamask or WalletConnect. Coinbase spokesperson McKenna Otterstedt said the company was launching an internal investigation into the site’s integration with the company’s wallet. Metamask is owned by Consensys, which said it had been unaware of the site before WIRED’s reporting and had launched an investigation.
The Scale Visible On The Site
WIRED monitored the site for eight days. During that period, five new images of women appeared on the Home feed and three appeared on the Explore page.
The site’s own stats showed that most of those images accumulated hundreds of “views.” WIRED said it was unclear whether every submitted image appeared on the feeds, or how the site counted views. Every post on the Home feed had at least a few dozen views.
The “Most Viewed” list was topped by celebrities and people with large Instagram followings. The most-viewed people of all time on the site were actor Jenna Ortega with more than 66,000 views, singer-songwriter Taylor Swift with more than 27,000 views, and an influencer and DJ from Malaysia with more than 26,000 views.
Swift and Ortega had already been targeted with deepfake nudes before. The circulation of fake nude images of Swift on X in January drew renewed attention to deepfakes and legal protections for victims. This month, NBC reported that Meta had hosted ads for a deepnude app for seven months, and that the app used a picture of Jenna Ortega from when she was 16 years old.
Why Victims May Never Know
The report places one site inside a larger pattern. OpenAI and Stability AI say their image generators are intended for commercial and artistic uses and include guardrails against harmful content. But WIRED notes that open source AI image-making technology is now relatively powerful, and that creating pornography is one of its most popular use cases.
Mary Anne Franks, a professor at the George Washington University School of Law who has studied nonconsensual explicit imagery, told WIRED that the site points to a wider hidden problem. Public cases tend to surface only when images are shared in a community and someone raises an alarm.
“There's gonna be all kinds of sites like this that are impossible to chase down, and most victims have no idea that this has happened to them until someone happens to flag it for them,” Franks says.
In the US, no federal law targets the distribution of fake, nonconsensual nude images, though a handful of states have enacted their own laws. AI-generated nude images of minors are treated differently by NCMEC.
“If it is indistinguishable from an image of a live victim, of a real child, then that is child sexual abuse material to us,” Newman says. “And we will treat it as such as we're processing our reports, as we're getting these reports out to law enforcement.”
Jennifer Newman, executive director of the NCMEC’s Exploited Children’s Division, said that in 2023, NCMEC received about 4,700 reports that “somehow connect to generative AI technology.”
The central warning is simple: deepfake nude generators do not only threaten public figures. They can turn casual selfies, classroom photos, social media posts and pictures of strangers into targets for nonconsensual abuse.