Hundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features

AI platform teams and app stores have a discovery problem: harmful generation tools can hide behind ordinary AI products. A WIRED investigation identified roughly 400 apps and websites linked to Chinese-language developers that appeared to offer mundane AI features while concealing tools for creating sexually abusive and violent deepfake images and videos. The findings suggest that reviewing an app's public interface, store listing, or default workflow is not enough. Builders distributing generative media products should audit hidden routes, remote feature flags, account-gated functions, and post-install updates, while marketplaces need ongoing behavioral testing rather than one-time approval. The scale of the network also points to coordinated reuse of infrastructure and code, making developer identity, payment links, domains, and shared backends useful signals for enforcement.