Apple rewrote parts of its App Store guidelines on Friday, and the changes go straight at two problems that have followed the store for years: AI apps that undress photos of real people, and apps that pass review looking like one thing and turn into another. The updated App Store guidelines appeared in a pair of developer posts dated October 16, and they close some gaps that San Francisco’s city attorney pressed on in July.
Nothing in the new App Store guidelines is a brand-new ban. Apple says as much in its own announcement. What changed is the wording, and the wording matters because it spells out what Apple will do to developers who break the rules.
What changed in the App Store guidelines
Apple’s companion post, titled “Strengthening App Review Guidelines against nonconsensual intimate imagery and bait-and-switch apps,” names three edits. Guideline 1.2, which covers user-generated content and moderation, now explicitly includes nonconsensual intimate imagery, including images made with AI tools. Apps must proactively prevent and remove that material. A developer who doesn’t put adequate safeguards in place, or who fails to fix violations after being told about them, risks breaking Apple’s Developer Code of Conduct.
A new Guideline 2.5.19 addresses hidden, dormant or undocumented features. The ban itself is old; it used to sit under Guideline 2.3.1. Apple pulled it out and gave it its own number, with a plain example: adding features that weren’t in the version submitted to App Review. Apps caught doing that can be removed from the App Store and from the Apple Developer Program.
Section 5, which deals with human trafficking and the exploitation of children, now says that apps facilitating either will be removed from the App Store and blocked from alternative distribution. Apple says it already removes such apps when it finds them, so the new sentence mostly makes the consequence explicit.
The full list of edits to the App Store guidelines in Apple’s separate announcement is longer, though most of it is housekeeping. A new Guideline 2.3.14 covers creative assets, and Apple adjusted 3.1.1(a) to fold a music-streaming bullet into its StoreKit external purchase link entitlements. Apple didn’t give an effective date in either post. It said translations of the guidelines would reach the Apple Developer website within a month.
Why nudify apps are the target
The phrase “nudification app” describes software that takes an ordinary photo and generates a sexually explicit version of the person in it, without that person’s knowledge. Pornography has been banned on the App Store for a long time, and Apple’s post says the ban already reaches apps that generate nudity or explicit imagery of real people. The trouble has been enforcement, and some of the apps present themselves as something harmless. The 13 apps San Francisco later named were marketed as face-swap tools.
That gap is what the City Attorney of San Francisco went after in July. According to a MacRumors report citing WIRED, the office sent cease-and-desist letters to Apple and Google on July 16 over 13 AI face-swap apps, eight on the App Store and five on Google Play. The letters demanded that both companies pull the apps, cut ties with the developers and stop taking a share of in-app payments, arguing that doing otherwise amounted to aiding and abetting the sale of explicit deepfakes. Apple told MacRumors at the time that it considered nudification apps a violation of its guidelines, had removed three of the named apps and was terminating their developers’ accounts. It said it was in contact with four others. Google told WIRED it had deleted hundreds of apps with nudifying features.
The pressure isn’t limited to small developers. MacRumors noted that the rewritten 1.2 explicitly names AI content-creation tools, and it suggested the language could reach large chatbots such as xAI’s Grok, which users had reported being used to make sexualized deepfakes. Apple reportedly told xAI earlier this year that Grok could be removed, and xAI added safeguards before Apple approved an updated version. That is MacRumors’ reading, and Apple’s own post doesn’t name any app. Grok was also the subject of our report on a New York City Council hearing about the chatbot.
The bait-and-switch problem
The 2.5.19 addition to the App Store guidelines matters beyond deepfakes. Apple says malicious developers, sometimes working in coordinated fraud operations, hide harmful content that only switches on after approval. A reviewer sees one product, and the harmful part activates only after the app is approved.
Apple put numbers on the scale. In its post, the company says it prevented more than 150,000 apps with safety issues from reaching customers last year and terminated more than 193,000 developer accounts for fraud and abuse. The post doesn’t break those figures down by guideline, so there’s no way to tell how many involved intimate imagery and how many were scams, copycats or malware.
The practical effect of giving the rule its own number is simple. A developer who ships a feature that was never disclosed to App Review can no longer argue that the policy was vague. Apple can point at 2.5.19 and cite the example it published.
What it means for developers
For developers, the biggest shift in the App Store guidelines is the word “proactively.” A moderation system that waits for user reports and then responds may not satisfy Guideline 1.2 for an app whose purpose is generating or editing images. Apple’s wording points toward filters, detection and removal workflows that run before a victim has to complain. Smaller studios building on top of image-generation models may feel that most, because the safeguards have to be built into the product and not bolted on afterward.
Developers who update apps remotely are in a second group to watch. Anything that changes what an app does after review, and likely that includes a feature switched on from a server, has to match what was disclosed during submission. Apple’s own example says features that weren’t in the submitted version are the problem, and the penalty listed is removal from the Developer Program, not just a rejected update.
What it means for users
For people who use iPhones, the App Store guidelines change is mostly invisible, which is the point. If the rules work, fewer of these apps survive in search results and fewer bait-and-switch updates reach devices. Rules don’t catch everything, though, and the July letters from San Francisco suggest enforcement has lagged behind policy.
People who find a nonconsensual explicit image of themselves or of a child have other routes besides the App Store. The Federal Trade Commission has published guidance on reporting platforms that don’t remove intimate images, and the Take It Down Act requires covered platforms to offer a removal process. Apple’s guideline change doesn’t replace any of that. Legal and financial pressure on consumer AI apps is also spreading well beyond the app stores, as our coverage of the Character.AI bankruptcy shows.
Editor’s take
Apple’s update to the App Store guidelines is a sensible tightening, but it’s a policy statement and not proof of enforcement. The test is whether the next set of nudify and bait-and-switch apps gets caught at review, or whether regulators and reporters keep finding them first. Watch for removals Apple announces on its own, not the ones it makes after a city attorney’s letter. If Apple can show that the proactive standard in 1.2 is being applied to the big AI products as well as the obscure ones, the rewrite will have teeth. If only small developers get removed, it will look like cleanup after the fact.
