Apple's new App Store rules target AI nudify apps and 'bait-and-switch' features

Apple rewrote its App Review Guidelines on Friday to bring AI image generators explicitly under its content-moderation rules, naming nonconsensual intimate imagery, including imagery made with AI tools, as something apps must proactively prevent and remove. The company published the update at 1:05 p.m. Pacific time, five minutes after a companion note stating that "nudification" apps, which create sexually explicit images of real people without consent, remain banned as pornography.

The second half of the change goes after how those apps get approved in the first place. Apple moved its rule against hidden features into a new guideline, 2.5.19, and added language that wasn't in the version TECHi compared from Oct. 11: attempting to hide undocumented features in an app binary, "such as through code obfuscation or other methods," is now named as grounds for removal and termination of the developer account. Apple called the tactic bait-and-switch, where an app shows App Review one thing and switches on a harmful capability after approval.

What Apple changed in guideline 1.2

The old guideline 1.2 was written for social networks. It opened with "apps with user-generated content present particular challenges, ranging from intellectual property infringement to anonymous bullying," and required filtering, a report mechanism, user blocking and published contact details from "apps with user-generated content or social networking services," according to the Oct. 11 copy of the guidelines preserved by the Internet Archive.

The new text widens the scope to "apps that enable users to create, generate, upload, share, or otherwise contribute content," names nonconsensual intimate imagery among the challenges, and states that "this includes social media, user-generated content platforms, and AI content creation apps," according to the current App Review Guidelines. The four required safeguards are unchanged. What changes is who they now plainly apply to: an app whose users type a prompt and receive an image is held to the same moderation duties as an app whose users upload a photo.

Apple also added a sentence that raises the stakes for developers who don't act. "Developers who fail to implement appropriate safeguards or fail to remediate identified violations after notice may be considered in violation of the Developer Code of Conduct," the guideline now says. The previous text described a sequence that ended at removal of the app. Tying the failure to the Code of Conduct puts the developer account, not just one app, in scope.

Why the bait-and-switch rule matters more than the ban

Nudification apps were already prohibited. Guideline 1.1.4 has long barred overtly sexual or pornographic material, and Apple's companion note says the ban "includes apps that generate nudity or sexually explicit imagery." The harder problem is detection. An app that submits a harmless photo editor for review and only later enables a nudify feature passes the check it was given.

That is what 2.5.19 addresses. The core sentence, "Don't include any hidden, dormant, or undocumented features in your app," previously sat inside guideline 2.3.1(a) next to rules about misleading marketing. Apple has now made it a standalone guideline and added the obfuscation clause, which did not appear anywhere in the Oct. 11 text by TECHi's search of both versions. The companion note describes the target as apps that "misrepresent their purpose during review and only reveal harmful capabilities — such as 'nudification' — after approval," and says apps found doing so face removal from the App Store and from the Apple Developer Program.

There is a limit worth noting. The new guideline text speaks about features hidden "in your app binary." A feature switched on later from a developer's server does not necessarily sit hidden in the submitted binary. Apple's companion note frames the rule more broadly, as covering apps that change "after approval" by adding features that weren't in the version submitted to review, and the existing 2.3.1(a) requirement to describe all new features in review notes still applies to every update. How Apple enforces server-side switches will show in removals, not in the rule text.

The alternative-distribution line in Section 5

The update also extends one enforcement step beyond the App Store. Section 5 previously said that, in extreme cases such as apps that facilitate human trafficking or the exploitation of children, "appropriate authorities will be notified." The new text adds that such apps "will be removed from the App Store and blocked from alternative distribution."

Apple already reserved a similar block for misleading marketing under 2.3.1(a), so the power isn't new. Writing it into Section 5 makes explicit that installing an app through an alternative marketplace or another distribution route outside the App Store is not a way around a removal for child exploitation or trafficking.

The purchase-link change developers will notice

One change is unrelated to safety. Guideline 3.1.1(a) used to let developers link to "a website the developer owns or maintains responsibility for" when they hold an external purchase link entitlement. It now says "a permitted destination," adds that "approved destinations may vary by region," and folds the separate music-streaming entitlement into the StoreKit external purchase link bullet. The United States storefront, where Apple already allows buttons and external links without an entitlement, is unchanged. TECHi covered the court fight behind that US exception when the Supreme Court declined to block the App Store ruling.

Apple also added guideline 2.3.14 on "creative assets," the new promotional images that can appear in product page headers and search results. They must relate to the app, be suitable for all audiences and avoid unverifiable claims or discounted prices. Translations of the revised guidelines will follow within a month, Apple said.

What would show the rules are working

Apple put a number on the scale of the problem in its note. "Last year, we prevented over 150K apps with safety issues from reaching customers and terminated over 193K developer accounts for engaging in fraud and abuse," the company wrote, adding that some violations come from coordinated fraud operations that conceal harmful content until after approval.

The strongest case against reading too much into Friday's update is that most of it restates bans Apple already had. Pornography was prohibited, hidden features were prohibited, and nudify apps were already out of bounds on paper. The test is whether the clearer scope and the Code of Conduct link change outcomes: fewer AI generators that produce sexual images of real people reaching the store, faster removal after notice, and terminations of the developers behind them rather than single apps. Deepfake image tools have already drawn government action, as TECHi reported when Grok's image features triggered bans and backlash earlier this year. Apple's next fraud-prevention report, and any named removals between now and then, are the evidence that would confirm or weaken the case that this update does more than tidy the rulebook.