Apple's Tougher AI Reviews

Apple is asking more artificial intelligence app developers to show exactly how their apps work before they reach the App Store, including screen recordings, clearer descriptions, and lists of outside services the apps call. (developer.apple.com; x.com) The new demands are showing up in App Review conversations inside App Store Connect, where Apple already tells developers to provide account details, required settings, and special instructions if reviewers need them to test an app. Apple says incomplete submissions can slow review or fail it. (developer.apple.com) Apple’s published rules already give reviewers broad grounds to reject apps for crashes, placeholder content, misleading metadata, or spammy duplication. The June 2025 App Review Guidelines list “App Completeness,” “Accurate Metadata,” and “Spam” as separate review categories. (developer.apple.com; developer.apple.com) That matters for AI apps because many of them are wrappers around the same models, with features that can be hard to verify from screenshots alone. Apple’s guidelines also single out chatbots under section 4.7, which covers mini apps, streaming games, chatbots, plug-ins, and emulators. (developer.apple.com) Apple has been tightening AI-related review rules in other places too. In November 2025, it updated the App Review Guidelines to require apps to disclose and get explicit permission before sharing personal data with third-party artificial intelligence services. (techcrunch.com; developer.apple.com) Developers describing the newer review requests said the target is low-quality or misleading submissions, not AI as a category. Their posts said teams are now packaging demos, reviewer notes, and backend service lists earlier in the release process. (x.com; x.com) Apple has long framed the App Store as a curated marketplace rather than an open software shelf. On its review page, the company says every app and update is reviewed to provide a “safe and trusted experience,” and that more than 40% of unresolved review issues involve completeness problems under guideline 2.1. (developer.apple.com; developer.apple.com) For AI app makers, that means the submission package is becoming part of the product. If reviewers cannot see the model’s behavior, the connected application programming interfaces, or the intended use case quickly, the app is less likely to move through review on the first pass. (developer.apple.com; x.com)