This article may contain affiliate links. Learn more on our Partner Program page.
AI apps that show real people naked without their consent have been a concern for Apple for months. Now the company has tightened its App Store review rules in three areas – and is also targeting a trick that such apps have previously used to pass the review process.
Apple updated its App Review Guidelines on October 16th. These are the rules by which every app is reviewed before being accepted into the App Store. In its notification to developers, Apple explicitly mentions two objectives: non-consensual intimate images, including those generated by AI tools, and so-called bait-and-switch apps, which only reveal their true function after approval. The full text of the rules can be found in the App Review Guidelines.
This issue has a history. In April, it was revealed that Apple had internally threatened xAI with the removal of Grok from the App Store after the chatbot generated sexualized images of real people. All news related to Apple's platform is compiled on the App Store topic page.
Key Facts at a Glance
- Rule 1.2 now explicitly requires that apps with user content prevent and remove non-consensual intimate images – including AI-generated ones.
- The new rule 2.5.19 prohibits the subsequent addition of features after release that were missing in the tested version.
- According to section 5, apps that facilitate human trafficking or the exploitation of children will also be blocked from alternative distribution channels in the future.
- Last year, Apple stated that it stopped more than 150,000 apps with security problems and terminated over 193,000 developer accounts.
What's changing in the rules?
Apple emphasizes that pornographic content has always been prohibited in the App Store, as have apps that generate nude images. The changes further tighten this policy in three areas.
| Rule | What's new |
|---|---|
| 1.2 (User content) | Moderation must explicitly prevent the collection of non-consensual intimate images, including AI-generated ones. Apps must actively prevent and remove them. |
| 2.5.19 (new) | No features will be allowed after release that were missing in the reviewed version. Violations will result in exclusion from the App Store and the Developer Program. |
| Section 5 (Legal) | Apps that facilitate human trafficking or the exploitation of children will be removed and blocked from alternative distribution channels as well. |
Rule 1.2 focuses on the responsibility of developers. According to Apple, those who fail to implement appropriate safeguards or do not address reported violations after being notified are in breach of the Developer Code of Conduct and risk having their app removed.
The trick with the added function
The new rule 2.5.19 targets a tactic that Apple itself describes: Apps appear harmless during the review process and only activate dangerous features, such as the artificial undressing of people, after they are available in the App Store. This was already prohibited in principle. Rule 2.3.1 prohibits hidden, dormant, or undocumented features; the new rule now specifically addresses this case.
Apple justifies the move by citing the methods of organized fraud networks that deliberately hide harmful content and only activate it after it has been approved. The statement reads:
We will not hesitate to remove apps and ban the developers behind them if they violate these guidelines.
What the ban on alternative distribution channels means
The change in Section 5 primarily affects markets where iOS apps are not exclusively distributed through the App Store. In the EU, apps may also be installed via alternative marketplaces and directly from the web. Previously, Section 5 stated that the relevant authorities would be notified in cases of human trafficking and child exploitation. Now, Apple clarifies that such apps will also be blocked through these channels.
Why the rules are only now following suit
I believe the changes are overdue, but fundamentally correct. The prohibition of nude image apps has long been in the rules – the problem was enforcement for apps that only reveal their functionality after review. With rule 2.5.19, Apple now has a specific basis for removing precisely these apps, along with their developer accounts.
For users, little will change immediately. What remains more important is the ability to identify fake images and voices, as the tools for doing so don't disappear with individual apps being removed from the store. How to detect AI fraud using deepfakes and fake voices is independent of Apple's rules.
Are stricter review rules enough to keep nude image apps out of the App Store – or should Apple be taking a closer look at AI apps even before they're released? Let us know in the comments where you would draw the line.
