APP STORE June 22, 2026 · 3 min read

App Store Review Guidelines for iOS 26: New AI Disclosure Rules and Minimum OS Targets

Apple updated its App Store Review Guidelines for iOS 26, adding explicit rules for AI-generated content and Foundation Models usage, while minimum deployment target pressure is building toward iOS 16. Here is the practical breakdown for your next submission.

By the AppsOps news desk · June 22, 2026 · Original source ↗

Apple shipped updates to its App Store Review Guidelines alongside the iOS 26 developer beta, adding clearer rules for apps that use AI-generated content or Apple's on-device Foundation Models framework, and tightening the language around minimum deployment targets. If you have a submission planned in the next 60 days, these are the changes worth reading before you hit upload.

AI and Foundation Models: What the New Rules Actually Say

iOS 26 ships with Apple's on-device Foundation Models framework — a set of APIs that give apps access to Apple's compressed language models without making any network calls. With that framework comes a new layer of App Review attention on any app that uses generative AI, whether on-device or via a third-party cloud service.

On-device Foundation Models (Apple's own APIs)

Apps using the official Foundation Models framework appear to face a lighter-touch review path: because Apple controls the model weights and outputs remain on-device, standard text-generation or summarisation use cases are unlikely to raise flags. That said, Apple's guidelines have long required that apps not position AI output as a substitute for professional advice — medical, legal, or financial. According to updated Apple developer documentation, that requirement is now stated more explicitly and applied to on-device generation too, not just cloud-based responses.

Third-party cloud AI (OpenAI, Anthropic, Gemini, and others)

Apps calling third-party LLM APIs are now expected to meet a higher transparency bar. Reports from WWDC 2026 sessions and Apple Developer Forums suggest reviewers are watching for three things:

A clear disclosure in the app description that AI-generated content is produced or used
An in-app mechanism for users to flag incorrect or harmful AI output
No use of AI to generate content that would itself violate the Review Guidelines — hateful, sexual, deceptive, or spam content

It is not yet clear whether Apple will codify these as a distinct guideline section or enforce them under the existing human-interface and metadata rules. What is clear from developer forum discussions is that AI wrapper apps — thin shells around a third-party model with no differentiated value — are receiving closer scrutiny, and the updated language gives reviewers a firmer basis for rejection.

If your app uses Apple's Foundation Models for on-device inference, see our earlier breakdown of what the Foundation Models framework actually offers for context on where the on-device/cloud line is drawn.

Minimum Deployment Target: The iOS 16 Cliff Is Approaching

Apple's pattern on minimum OS targets is consistent: within 12–18 months of a major iOS release, new submissions start seeing friction if they target an OS more than three major versions old. With iOS 26 in beta, that pressure is now pointing at apps still targeting iOS 15 or below.

No hard cutoff has been announced yet, but the signals are converging:

WWDC 2026 code samples consistently used iOS 16 as the floor — iOS 15 examples were absent from new-framework sessions
StoreKit updates announced at WWDC — improved win-back offer APIs and a subscription analytics endpoint — require iOS 16 at minimum
App Review feedback on unrelated rejections has reportedly started citing minimum OS version as an ancillary note, a pattern the community has flagged on Apple Developer Forums

For most apps, iOS 14 and 15 combined user share is now well under 5%. If your analytics confirm that, the case for delaying the minimum-target bump weakens considerably. The unlock you get — access to newer StoreKit subscription APIs, improved in-app purchase analytics, and smoother App Store Connect reporting — outweighs the marginal user loss in most categories.

Practical Checklist for Your Next Submission

Scenario	Action Before Submitting
App uses Apple Foundation Models	Confirm app description reflects AI use; avoid positioning output as professional advice
App calls a third-party LLM API	Add in-app disclosure and a user-feedback mechanism for AI output
App targets iOS 15 or below	Pull App Analytics — if sub-5% on iOS 15, plan the minimum-target bump in your next release
App uses AI to generate paywall or subscription copy	Human-review all generated strings; Apple has rejected apps for deceptive pricing claims, and unreviewed AI copy adds risk
App ships AI-translated App Store metadata	Ensure human review of all localized descriptions — mistranslated AI output in a secondary language is a fast path to rejection

That last row is worth singling out: if you are using AI-assisted translation to push your App Store metadata into multiple languages, the translated output needs human review before submission. Apple's guidelines have always required metadata accuracy across all storefronts, and a flagged mistranslation in a secondary territory can hold up your entire release. See how localized App Store screenshots and metadata are handled at scale, or check the pricing overview if you are evaluating the cost of a full localization pass before an iOS 26 launch.