App Store AI Surge & Friction

For years, the App Store looked mature in the least exciting way. Fewer new apps arrived each year. Then AI coding tools hit the market and the curve bent upward almost at once. In the first quarter of 2026, Apple’s App Store logged 235,800 new apps, up 84 percent from a year earlier, according to data from Sensor Tower reported by The Information and picked up widely on April 6. The point is not just that more people are coding. It is that far more people can now ship something that looks like an app at all (9to5mac.com, gizmodo.com). That matters because Apple’s store was built for a slower world. The App Review system assumes an app is a finished object. A developer submits a bundle. Apple checks it. Users download that bundle. AI coding tools break that model by making software feel fluid. A prompt can generate a new feature, a new interface, or a new mini-app after review is over. Apple’s own guidelines still say apps must be “self-contained” and may not download, install, or execute code that changes features or functionality (developer.apple.com, apple.gadgethacks.com). That rule, Guideline 2.5.2, is the hinge of this story. In March, Apple blocked updates to AI-assisted coding apps including Replit and Vibecode, and it pulled at least one other “vibe coding” app, Anything, from the store. The issue was not that AI helped build the apps. Apple has said it has no rule against AI-built apps as a category. The issue was that these products let users generate and run fresh software inside the host app itself. That is the old “app within an app” problem in new clothes (9to5mac.com, apple.gadgethacks.com, appleinsider.com). Once you see that, the apparent contradiction disappears. Apple is happy to sell developers better tools. It is much less happy to allow apps that become development platforms on the device. That distinction is why Apple can embrace AI features in Xcode while rejecting third-party apps that turn an iPhone into a live code execution environment. The company is not drawing a line between human-written software and AI-written software. It is drawing a line between reviewed software and software that keeps rewriting itself after review (developer.apple.com, medianama.com). Developers are now testing that line in court. Gadget Hacks reported that a developer whose AI coding app was removed has sued Apple over the takedown tied to Guideline 2.5.2. The problem for that plaintiff is that Apple’s contracts give the company broad power over distribution, and a federal judge recently dismissed a similar challenge from the music app Musi after finding Apple could remove an app with notice, even without cause. That does not make Apple’s policy elegant. It does show how much control is embedded in the App Store’s legal structure, not just its technical rules (apple.gadgethacks.com, developer.apple.com). The legal pressure is coming from the other side too. On April 6, Apple moved to take its fight with Epic Games back toward the Supreme Court, asking for review of the rulings that curtailed its ability to impose a 27 percent commission on purchases made through external payment links. The Ninth Circuit had already upheld the contempt finding against Apple in late 2025, and Apple’s rehearing bid was denied in March 2026. So at the same moment Apple is trying to contain AI-generated software inside the store, it is also trying to preserve its power to charge for commerce that happens outside it (techcrunch.com). That is why the App Store suddenly feels unstable. AI has made app creation cheaper and faster. Apple’s rules were designed to keep apps fixed, inspectable, and monetized on Apple’s terms. Those two facts can coexist for a while. They get harder to reconcile when 235,800 new apps show up in a single quarter and some of the most ambitious ones are trying to build the next app after the user opens the first one (9to5mac.com, developer.apple.com).