Apple’s app store chokes on ai-generated apps as review queue hits one-week delays
Tim Cook’s walled garden is wilting under its own perfectionism. Overnight, the App Store’s human-only review pipeline buckled under an avalanche of vibe-coded throwaways, pushing average approval times from 24 hours to three days and, in scattered cases, a full week—an eternity in mobile commerce.
Manual gatekeepers drown in synthetic code
Developers who once hit “submit” on a Thursday and woke to sales charts on Friday now refresh iTunes Connect like lottery junkies. The spike is not speculative: 9to5Mac’s internal metrics show daily submissions have doubled since December, when ChatGPT-coded templates began circulating on Reddit and TikTok. Apple’s 500-plus app reviewers, each expected to probe for hidden APIs, copy-cat clones and data-grabs, now confront binaries that were literally written by a language model in minutes and debugged by nobody.
The economics look paradoxical. More apps equal more 30 % cuts for Apple, yet only if the queue moves. Every stalled approval is a potential Fortnite-scale revolt in miniature; indie devs already whisper about shifting paid apps to progressive web apps that bypass Cupertino entirely. One-week limbo can kill a marketing window—ask any studio that buys TikTok trends by the hour.
Why the bots beat the humans
Vibe coding’s seductive promise—plain-English prompts to full Xcode projects—erases the friction that once separated “idea” from “IPA file.” The result is software that compiles, even runs, yet carries the DNA of statistical hallucination: hard-coded keys pasted from Stack Overflow answers that never existed, privacy labels copy-pasted from yoga apps now grafted onto crypto wallets, entire Swift files that reference Android classes. A human reviewer must untangle this spaghetti line-by-line, because Apple’s static analyzer was built to flag known malware patterns, not the surrealist edge cases of generative AI.
Cupertino’s refusal to deploy large-scale automation is ideological. The company markets App Store safety as handcrafted, artisanal, the digital equivalent of a Kyoto knife maker. That story collapses when a 19-year-old in Jakarta can ship 47 flashlight utilities in a single afternoon, each embedded with a different hallucinated SDK.
The coming fracture
Apple will not hold the line forever. Insiders tell me the review division has quietly requested budget for “intelligent pre-filtering,” code for ML models that pre-score submissions before human eyes land on them. The shift would mirror the 2016 pivot when Apple first let algorithms suggest editorial content on the Today tab—heretical at the time, standard now.
Expect a tiered system by autumn: fast-track green-lane for developers with spotless five-year records, and a probabilistic auto-scan for the long tail, with random human audits to keep the PR department happy. The 30 % commission will stay; the fairy tale that every app is cradled by a caring expert will not.
Until then, the countdown is raw. Developers are already stockpiling binaries like canned food, guessing which week Apple will break its own embargo and admit the bots have won. When that press release drops, it won’t be titled “We surrendered.” It will be called “Enhanced review acceleration tools”—and the footnote will reveal the median review time never returned to 24 hours again.