Apple Makes AI Image Moderation a Condition for Apps

Apple revised its App Review Guidelines over the weekend, and the headline change is aimed squarely at apps that generate images. Under an update published on October 16, the company wrote nonconsensual intimate imagery into the moderation duties that every app hosting user content already carries, explicitly including imagery produced with AI tools. Apps that fail to build safeguards, or that do not fix violations after being told, now risk a Developer Code of Conduct violation rather than a routine rejection.

The revision touches three sections of the guidelines. Section 1.2 was clarified so that content moderation requirements cover nonconsensual intimate imagery, AI-generated or otherwise, and require developers to proactively prevent and remove it. Section 2.5.19 is a new numbered rule that formalises a long-standing prohibition on hidden, dormant, or undocumented features, aimed at apps that present themselves as something harmless during review and then switch once approved. Section 5 was tightened so that apps facilitating human trafficking or child exploitation are removed and blocked from alternative distribution, not merely pulled from the store.

What actually changed

The bait-and-switch clause is the one that should worry engineering teams. Apple's wording covers apps that misrepresent their purpose during review and "only reveal harmful capabilities" after approval, and it removes the grey area around shipping capability behind a server-side flag. Apple now says those apps face removal from the App Store and from the Apple Developer Program, which means the organisation loses its account, not just a binary.

Why Apple moved now

In July, San Francisco City Attorney David Chiu ordered both Apple and Google to clean up and delete nudification apps, accusing the companies of profiting from them through store fees. An April Tech Transparency Project report had already found multiple nudify apps on both stores, some rated for minors.

What it means for builders

For anyone shipping generative features on iOS, moderation stops being a launch-week checkbox and becomes infrastructure. That means a filtering layer in front of generated output, in-app reporting, and account blocking, all present in the build Apple reviews rather than added later. It also means the capability that ships in the reviewed binary has to match the capability the product can actually deliver, which is a real constraint for teams that stage AI features behind flags. The safer pattern is to ship the moderation, disclosure, and reporting surfaces with the feature flag itself, so the reviewed build already contains the controls.

Google Play has not published an equivalent rule change this week, but the San Francisco order named both stores, which makes it reasonable to expect pressure there too.