Hundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features

At first glance, the self-proclaimed “pet portrait” app Viefur appears to be just another AI slop generator. Using digital coins sold in the app—around $4 will buy 130 coins—pet photos can be turned into an “anthropomorphic portrait,” with the critter dressed in “human clothing” and inserted into bizarre scenes. Think a hamster wearing beach shorts, sunglasses, and lounging in a swimming pool or a Hawaiian shirt-wearing cat, drinking from a coconut.

All of this is cover for what appears to be the app’s true purpose: creating sexually abusive deepfake images.

A WIRED investigation found the app is part of a network of more than 400 connected domains and apps that, like Viefur, often claim to be everyday services—such as language tuition or AI workout coaches. When accessed under certain conditions, however, apps in the network stealthily load a web interface allowing the generation of explicit AI sexual images and videos using photos users upload. The apps include AI scenes such as “bedroom rape” and “threatening knife,” where a man holds a knife to a woman’s neck.

The network was discovered hiding in plain sight, by multiple researchers, after they investigated links from thousands of video ads containing explicit deepfake content that were published on Meta’s Facebook and Instagram. As WIRED has reported in recent months, hundreds of these ads contained AI generated child sexual abuse material, including images of real children and a member of a European royal family. The ads reached tens of thousands of people and have prompted investigations from regulators around the world.

In each case, the ads linked out to the deceptive apps in Apple and Google’s app stores—allowing people to generate explicit images and make payments using Apple and Google Pay. The network behind the nudify apps appears to have been publishing them for around a year, at times publishing dozens per week. This is despite Google, Apple, and Meta saying they enforce their policies that do not allow harmful deepfake content on their platforms.

The apps and domains investigated by WIRED appear to be run by Chinese developers and a Hong Kong registered entity called KeLing Technology Limited, according to a review of web domains and how the services operate. Email addresses linked to the entity, apps, and related services, did not respond to multiple requests for comment from WIRED, but after being contacted, they appeared to erase public connections between KeLing Technology Limited and the deepfake tools.

“This is an organized, commercially run operation: One operator behind hundreds of disposable apps and domains, cloaked to pass App Store review and pushed through tens of thousands of Meta ads under fake payer names,” says Paul Bouchaud, a lead researcher at the nonprofit AI Forensics, who investigated the network. “They don’t expect enforcement to catch up, and so far it hasn’t.”

Hidden Behavior

Download one of the apps in the deepfake network and you may see exactly what was advertised—such as an app to create AI pet pictures. But behind the scenes, researchers found, the apps are quietly making determinations about whether to show the benign content or load a web interface showing explicit deepfakes.

“Each time the application is downloaded, there is a call to the server with some information, such as where the user is coming from, the language, some unique identifier, and then the server replies whether or not to show the explicit or the non-explicit version of it,” explains Bouchaud, who reverse engineered the apps’ and associated websites’ code. They say there’s no clear list of reasons why one user may be shown the explicit version of an app, but one trigger appears to be downloading it after clicking an associated ad on Meta’s platforms.