A Guideline 4.2 rejection for an AI app usually means Apple looked at it and saw a chat box sitting on top of an API, with nothing the web or a generic assistant does not already do. It is the modern version of the web-view rejection. The fix is not a prettier screen; it is giving the app a reason to exist as an app. Here is what 4.2 asks for and how to get there.
Short answer
Guideline 4.2, minimum functionality, rejects apps that do not go beyond a repackaged website or a thin wrapper around an API. Per Apple's guideline, an app should include features, content, and a user interface that raise it above web content and offer lasting value, or it does not belong on the App Store. For an AI app this hits a bare prompt field placed over a model API. The fix is to add native functionality and a focused purpose the web and a generic chatbot do not provide, not just a nicer interface around the same chat.
What you should know
- A chat box over an API is the trap: that alone reads as a thin wrapper under 4.2.
- Apple wants app-like value: features, content, and a UI that go beyond a website or a generic bot.
- Working is not enough: an app that functions can still lack the lasting utility 4.2 requires.
- It is the new web-view rejection: 4.2 and 4.2.3 cover both literal web views and thin API fronts.
- The fix is functional, not cosmetic: add what only a native app can do, not just styling.
What does Guideline 4.2 actually require?
That your app be useful, unique, and app-like rather than a repackaging of something already on the web. The guideline asks for features, content, and a user interface that go beyond a website, and for lasting entertainment value or adequate utility, and it states plainly that an app that lacks these may not be accepted. Two related clauses sharpen it: 4.2.2 says an app should not primarily be marketing material, web clippings, a content aggregator, or a collection of links, and 4.2.3 says the app should work on its own rather than deferring all of its function to downloaded web content. The common thread is that the value has to live in the app.
Why do AI chatbot apps get rejected under 4.2?
Because a thin chat interface offers nothing the user could not already get. If your app is a text field that forwards prompts to a model API and shows the reply, a reviewer sees the same experience as a website or a generic assistant, with a wrapper around it. There is no native capability, no specific problem it solves better than the open tools, and no reason it needs to be a download. Apple treats that as minimum functionality, the same way it treats a web view of your homepage. The model being impressive does not help, because the app is not adding anything to it.
What turns a wrapper into an app?
Native capability and a focused purpose. The difference is whether the app does something the web and a generic bot do not. The table contrasts the two.
| Thin wrapper, rejected | App-like, accepted |
|---|---|
| A chat box that forwards prompts to an API | Native features built around the AI, such as device integrations or saved, searchable history |
| The same value as the website or a generic assistant | A specific job the app does better than the open tools |
| A web view of your site | Native interface, offline behavior, and system integration |
| A collection of prompts or links | A focused tool with lasting, repeatable utility |
The right-hand column is not about more features for their own sake; it is about a clear use case that justifies installing an app.
How do you fix a 4.2 wrapper rejection, step by step?
Give the app native value, then resubmit:
- Name the thin part honestly: if the core is a prompt in and a response out, that is what the reviewer flagged.
- Pick a focused job: choose one task your app does better than a generic assistant, rather than being a general chatbot.
- Add native capability: integrate device features, store and search past results, support offline use, connect to a real workflow, or process the user's own files or data.
- Make it work on its own: ensure the app functions without redirecting to a website, per 4.2.3.
- Rebuild and resubmit, and in the review notes point to the native features that distinguish it from the web.
What to watch out for
The first trap is believing a polished interface fixes 4.2; styling does not add functionality, and reviewers look for capability, not decoration. The second is the general-assistant framing, since the broader and more generic the app, the more it looks like the tools that already exist; a narrow, specific app is easier to justify. Follow Apple's interface guidance for the app-like feel, but treat it as the finish on real functionality, not a substitute. Guideline 4.2 is a design and value judgment rather than a security finding, so it is separate from a pre-submission scan; a scan such as PTKD.com (https://ptkd.com) checks the binary against OWASP MASVS, while clearing 4.2 is about the functionality you add.
What to take away
- Guideline 4.2 rejects AI apps that are a thin chat box over an API, with no value beyond the web or a generic assistant.
- The fix is native functionality and a focused purpose, not a nicer interface around the same chat.
- Add device integration, saved and searchable history, offline behavior, or a real workflow, and make the app work on its own per 4.2.3.
- The 4.2 fix is about the app's functionality and is separate from the binary checks a pre-submission scan such as PTKD.com performs.
