Apple's New App Store Rules: What They Mean for AI Nudify Apps

By Jaspal Singh October 17, 2026 Updated: October 17, 2026

Apple's revised App Review Guidelines, published on October 16, 2026, make apps that create or share content responsible for preventing and removing nonconsensual intimate imagery, including AI-generated nudes. A new rule, 2.5.19, bans hidden or dormant features, which targets "bait-and-switch" apps that pass review as one thing and later turn into something else.

What Happened

On Friday, October 16, 2026, Apple posted two notices to its developer news feed. The first, "Strengthening App Review Guidelines against nonconsensual intimate imagery and bait-and-switch apps", explains the policy reasoning. The second, "Updated App Review Guidelines are now available", lists every section that changed. The revised App Review Guidelines were live the same day, with translations promised within a month.

According to Apple's change log, the edits touch six places:

Guideline 1.2 (User-Generated Content) now names nonconsensual intimate imagery as one of the core risks for apps that let people "create, generate, upload, share, or otherwise contribute content," and it explicitly lists AI content creation apps alongside social media and user-generated content platforms. Failing to put safeguards in place, or failing to fix violations after Apple gives notice, can now be treated as a breach of the Developer Code of Conduct.
Guideline 2.5.19 is new. It bans "hidden, dormant, or undocumented features" and says that hiding them in the app binary, "such as through code obfuscation," is grounds for removal and termination of the developer account. Apple says this restates a long-standing policy that used to live inside Guideline 2.3.1.
Guideline 2.3.14 is new and covers the creative assets developers can now show on product pages and in search results.
Guideline 3.1.1(a) folds the Music Streaming Services Entitlement into the StoreKit External Purchase Link bullet and clarifies that such links must go to a "permitted destination."
Section 5 (Legal) now says apps found to facilitate human trafficking or the exploitation of children will be removed from the App Store and blocked from alternative distribution.
The "Before You Submit" and "After You Submit" sections gained links to extra documentation.

Apple's policy post adds two figures: over the past year it says it stopped more than 150,000 apps with safety issues from reaching customers and terminated more than 193,000 developer accounts for fraud and abuse. Pornographic material was already banned under Guideline 1.1.4; the new language makes clear that this extends to apps that generate nudity, including so-called nudification apps.

Why It Matters

The update closes the gap between what Apple's rules said and what has actually been reaching the App Store this year. As 9to5Mac's report recounts, the Tech Transparency Project found dozens of AI "nudify" apps on the store in January 2026, and a follow-up in April found that App Store search suggestions and ads were steering people toward them. In July, San Francisco officials ordered Apple to remove eight apps capable of producing nonconsensual intimate images. The same month, 9to5Mac found apps disguised as games or utilities that revealed betting features only when opened from Brazilian IP addresses.

The other pressure point was xAI's Grok. Users generated sexualised images of women and minors with it earlier this year, and MacRumors reports that Apple warned xAI in April that Grok could be pulled unless it stopped allowing nude deepfakes. xAI added safeguards and Apple approved an updated version. Apple's new text does not name Grok or any other app.

For ordinary iPhone owners, the practical effect is that Apple now has a written basis to remove an app because of what its users or its AI model produce, not only because of what the developer shipped. For developers, it shifts the burden: an image generator, chat app or photo editor is expected to stop abuse before it happens, and to act quickly when Apple points out a problem.

How It Works

The changes work through three separate mechanisms, and it helps to look at each one on its own.

1. Moderation duties now cover AI output

Guideline 1.2 has long required apps with user-generated content to offer four safeguards: a way to filter objectionable material before it is posted, a way to report offensive content with "timely responses," the ability to block abusive users, and published contact details. What changed is the scope. By naming AI content creation apps and nonconsensual intimate imagery in the opening paragraph, Apple treats a picture produced by a model the same way it treats a picture uploaded by a user. If an app's generator can be prompted into creating a sexualised image of a real person, the filter requirement applies to that output.

Apple's policy post goes a step further than the guideline wording and says developers must proactively prevent and remove this material. In practice, that points to prompt filtering, classifiers that check generated images before they are shown, and reporting tools that reach a human. The guideline does not prescribe a specific technology.

2. The Code of Conduct becomes the enforcement lever

The second mechanism is the link to the Developer Code of Conduct. Before this update, a weak moderation system was mostly a reason to reject a build. Now, Apple says a developer who fails to implement appropriate safeguards, or does not fix violations after notice, may be violating the Code of Conduct itself. That matters because Code of Conduct breaches can cost a developer their whole account, not just a single app. Apple's post says plainly that it will "remove apps, and terminate the developers behind them."

3. Rule 2.5.19 targets the review loophole

Bait-and-switch apps exploit a simple weakness in any review process: reviewers can only judge the build in front of them. A developer submits a harmless photo filter, gets approved, and then switches on a nudification feature through a server flag, a region check or a later update. The Brazilian gambling apps worked the same way, showing betting screens only to certain IP addresses.

Guideline 2.3.1 already required developers to describe new features "with specificity" in their review notes. The new 2.5.19 turns the underlying principle into a standalone rule: no hidden, dormant or undocumented features, and functionality "should be clear to end users and App Review." Code obfuscation used to conceal a feature is named as grounds for removal and account termination. In other words, the offence is no longer just misleading the reviewer; it is shipping code whose purpose is concealed.

4. Alternative distribution is no longer an escape route

In the EU, iPhone apps can be installed from alternative marketplaces, and Apple still notarises them. The Section 5 change means an app pulled for facilitating human trafficking or child exploitation, including child sexual abuse material, cannot simply move to another marketplace. Apple will block it from alternative distribution as well and says appropriate authorities will be notified.

What's Still Unknown

Several important details are not in Apple's posts. There is no definition of "appropriate safeguards," so it is unclear how strong a filter must be, or whether a small developer using a third-party model API is judged by the same bar as a large platform. Apple also has not said how it will detect dormant features at scale, whether through static analysis of binaries, runtime testing from different regions, or user reports.

Apple has not said whether any apps were removed alongside the update, or how the rule applies to large general-purpose apps whose AI features can be misused, such as X with Grok built in. It is also unclear how the "after notice" window works: how long a developer has to respond before an account is at risk. Finally, the 150,000 and 193,000 figures cover all safety and fraud enforcement, not nudification apps specifically, so they do not show how big that problem is.

Frequently Asked Questions

When did Apple's new App Store rules take effect?

Apple published the revised App Review Guidelines on October 16, 2026, and the updated text was live on its developer site the same day. Apple did not announce a separate grace period, so the rules apply to new submissions and to apps already on the store. Translations of the guidelines into other languages are due within one month of the update.

What is a nudification app?

A nudification app uses generative AI to turn an ordinary photo of a person into a fake nude or sexually explicit image, usually without that person's consent. Apple already banned pornographic material under Guideline 1.1.4, and the update makes clear the ban covers apps that generate such imagery. Apps that allow this content to be created now face removal and possible developer account termination.

What does Guideline 2.5.19 actually ban?

Guideline 2.5.19 bans hidden, dormant or undocumented features. Everything an app does must be clear to its users and to App Review. A developer who hides functionality, for example behind code obfuscation, a remote switch or a region check that reveals features only in some countries, risks having the app removed and the developer account terminated. Apple says this formalises policy that sat in Guideline 2.3.1.

Does this affect AI image generators that are not nudification apps?

Yes. Guideline 1.2 now names AI content creation apps directly, so any app that generates images must filter objectionable output, let users report content, block abusive users and publish contact details. Apple's post says these apps must proactively prevent and remove nonconsensual intimate imagery. A general-purpose generator that can be easily prompted into making explicit fakes of real people could fall foul of the rule.

Will Apple remove Grok or the X app?

Apple has not said so. Its posts do not name any app. MacRumors reports that Apple warned xAI in April 2026 that Grok could be removed unless it stopped allowing nude deepfakes, and that Apple approved an updated version after xAI added safeguards. Under the new wording, any app with AI image tools that keep producing such content after notice could face action.

Can a removed app still be installed from an alternative marketplace?

For the most serious cases, no. Section 5 now says apps found to facilitate human trafficking or child exploitation will be removed from the App Store and blocked from alternative distribution, which applies to marketplaces Apple must allow in the EU. For other violations, such as weak moderation, Apple's posts only spell out removal from the App Store and possible developer account termination.

What should developers of user-generated content apps do now?

Review the four Guideline 1.2 safeguards against how the app actually behaves: filtering before content is posted, reporting with timely responses, user blocking and visible contact details. Apps with generative features should test whether prompts can produce explicit images of real people. Document every feature in the App Review notes, and remove any remote flags or region-gated behaviour that reviewers cannot see.