Hundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features
A few hundred apps sounds like a lot, but I'd bet this is only the visible crust of the problem. Anyone who has watched the deepfake space knows the tools move faster than the takedown notices, and the fact that these were hiding inside ordinary apps means users were likely finding them through word of mouth or encrypted channels, not through the store listing. The mundane wrapper is the smart part: it evades the filters that would catch an app proudly announcing itself as a deepfake generator.
The Chinese-language developer angle is worth sitting with, but I'm careful not to let it become the whole story. Enforcement in one jurisdiction tends to just push the operation somewhere else, so singling out one source risks false comfort. What I actually want to know is how these apps got distributed and why the stores did not catch them sooner, and whether any of the imagery generation was tied to real identifiable people, because that changes the legal stakes entirely.
The useful takeaway for me is less about the specific apps and more about the broader audit problem. At this point any AI tool that claims to do something benign deserves a closer look at what it can do off-script. Trusting a surface description is no longer a reasonable default.