Over 400 AI Apps Slip Past Apple and Google Vetting, Concealing Deepfake Abuse Tools

More than 400 interconnected apps and websites that quietly generate sexually abusive deepfakes slipped through the review systems of Apple and Google and were with ads on Meta's platforms, according to an investigation by WIRED.

The network, which researchers say appears to be run by Chinese developers and a Hong Kong-registered company called KeLing Technology Limited, has been operating for roughly a year. Apps in it present themselves as ordinary tools, such as pet-portrait generators, language tutors, or AI fitness coaches. Under certain conditions, however, they load a hidden interface that produces explicit imagery and video from photos uploaded by users, including scenes labeled "bedroom rape" and "threatening knife."

The findings raise fresh questions the ability of the largest technology platforms to police content on their own marketplaces, and how quickly operators can rebuild after enforcement actions.

A bait-and-switch built to pass review

According to the investigation, the apps behave normally when reviewers examine them. Each time an app is opened, it contacts a server with details such as the user's location, language, and a unique identifier, and the server decides whether to display the harmless version or the explicit one, said Paul Bouchaud, a lead researcher at the nonprofit AI Forensics who reverse-engineered the code. One apparent trigger for the explicit version is installing the app after clicking an associated ad on or Instagram.

The hidden layer lives on the apps' websites, which typically show only a blank black screen. Behind that screen sit dozens of pornographic templates into which users can insert photos of real people, along with a system for buying credits to generate images. Researchers at the security firm Infoblox identified nearly 500 linked domains that the same code and reference one another, with configuration files exposing encryption keys and advertising trackers.

Apple removed 19 apps, including a pet-photo app called Viefur, after the investigation shared details of the network. Adam Dema, an Apple spokesperson, said the company reviews every app before publication and that these apps complied with its rules at the time. "After approval, their developers changed the website the app pulls its content from, replacing what we approved with pornographic content, and it through pornographic ads hosted on a third-party platform," he said. Apple said it has suspended tens of thousands of apps over the past year for similar tactics and recently tightened its rules on so-called nudify tools.

Google removed most of the apps before being contacted, and pulled one of the most popular, DiffAI, shortly beforehand. DiffAI had been downloaded more than 100,000 times from the Play Store. Google did not respond to a request for .

Scale of the operation

Bouchaud's review of domain registrations found the apps began appearing in late 2025 and surged this spring, at one point reaching 30 new app-and-website pairs per week. In September, 27 were live on Google's store, and the apps had been downloaded more than 350,000 times in total, many of them rated for all age groups.

Note: Figures come from the investigation and from company statements.

Meta said it banned KeLing Technology Limited from advertising in 2025 after the company failed to comply with a cease-and-desist letter. After being shown the domains, a spokesperson said the company had reviewed more than 3,000 associated ads, two-thirds of which it had already removed, and had since taken action against 11,600 ad accounts, more than 4,700 business pages, 5,200 users, and 15,000 pages, while blocking 209 apps from advertising.

"The criminals behind them take increasingly deceptive steps to evade detection, hiding behind fake names and identities, running ads that look benign, and sometimes linking to Google's and Apple's app stores, with the ads' true intent revealed only after a click," the spokesperson said. The ads used apparently fabricated payer names and pages with no followers.

Chinese origins and a Hong Kong entity

Investigators traced the network to Chinese developers whose identities appear deliberately obscured. Maël Le Touz, a staff threat researcher at Infoblox, said the operation uses Chinese-only hosting, code written with Chinese-language , and promotional videos bearing watermarks from Chinese video-editing software. Privacy policies on the apps are generic, addresses are free Gmail or Outlook accounts, and developer names are obscure.

KeLing Technology Limited sits near the center of the operation, according to the investigation. Its name appeared as the for the first website registered in the network, and its brand name showed up in details embedded in an explicit template used by the apps. An administrative login page for the hidden interface referenced "Keling," and email addresses on the company's domain appeared on at least one app later removed from the stores. A newer AI model inside the interface, called Veyra, was linked on its website to a KeLing address.

Emails to KeLing addresses and to addresses listed in the apps' privacy policies went unanswered. After the inquiries, some associated social media channels were renamed, KeLing addresses were stripped from the Veyra site, and the company's address was removed from the first registered website.

"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," Bouchaud said. "They don't expect enforcement to catch up, and so far it hasn't."

Katie Paul, director of the Tech Transparency Project, which first surfaced ads depicting explicit deepfakes of minors, said the ads violate several of Meta's policies at once and that the platforms need to answer where the money from such apps flows and how their operators keep evading moderation.

What it means for the platforms

The case highlights a structural weakness in app-store vetting: review is a point-in-time check, while an app's behavior can be changed remotely after approval. Apple's account of the scheme, in which developers swapped the content source behind an approved app, describes exactly the gap that server-side controls create. Both Apple and Google have tightened policies on nudification tools, but the network's longevity and its ability to keep registering new domains suggest that takedowns have so far removed individual apps rather than the underlying operation.

For investors, the immediate financial exposure is limited, but the regulatory risk is not. Ad-dependent platforms face mounting scrutiny over illegal content in paid promotions, and app marketplaces face questions whether review and monitoring are adequate. The investigation notes that similar ads and apps were still being published as reporting concluded.