Widespread network of AI apps disguised as benign tools to generate explicit deepfakes

A comprehensive investigation uncovered a vast network comprising over 400 domains and applications, including seemingly innocuous services like pet portrait generators such as Viefur. These apps, which appear to be run by Chinese developers and a Hong Kong entity, stealthily load hidden web interfaces allowing users to create explicit AI images and videos, including scenarios like bedroom rape. Researchers found the operation was promoted through tens of thousands of Meta ads, some containing AI-generated child sexual abuse material, reaching tens of thousands of people. Paul Bouchaud, a lead researcher at AI Forensics, described it as an organized, commercially run operation, noting that the apps started appearing toward the end of 2025, with a spike in spring 2026. While Google and Apple removed numerous apps, Apple spokesperson Adam Dema stated that the apps complied with review rules before developers altered the content source to include pornographic material. Meta later reported taking action against 11,600 ad accounts and blocking 209 apps after being provided with the list of domains.