For years, getting an app into Apple’s App Store meant clearing a gauntlet of human reviewers, automated scans, and design guidelines strict enough to make a graphic designer weep. The process was slow. It was frustrating. But it worked — mostly — as a quality filter that kept the worst software from reaching a billion iPhones.
Now that filter is about to be tested like never before.
A new wave of so-called “vibe-coded” applications — software built largely or entirely by artificial intelligence with minimal human programming — is threatening to overwhelm Apple’s review infrastructure, according to developers and industry analysts tracking the trend. The concern isn’t hypothetical. It’s already starting to show up in longer review times, inconsistent feedback from Apple’s team, and a growing backlog that some developers say could get dramatically worse before it gets better.
Business Insider reported that developers are warning of a flood of AI-generated app submissions that could significantly slow Apple’s approval pipeline. The term “vibe coding” — coined by AI researcher Andrej Karpathy in early 2025 — describes a process where a user describes what they want an app to do in plain English, and an AI model like Claude, GPT-4, or one of several open-source alternatives writes the actual code. The human doesn’t need to understand the code. They just need to describe the vibe.
And it’s catching on fast.
Tools like Replit, Cursor, Bolt, and Lovable have made it possible for people with zero programming experience to produce functional mobile applications in hours rather than months. What once required a team of developers, a UX designer, and a project manager can now be approximated — sometimes convincingly — by a single person with a laptop and an AI subscription. The democratization of software development sounds inspiring in a keynote presentation. In practice, it means Apple’s review team is staring down a tsunami of submissions from creators who may not understand what their own code does.
The Scale Problem Apple Didn’t Plan For
Apple processes roughly 100,000 app submissions per week, according to figures the company has shared publicly. Its review team — a mix of human reviewers and automated tools — typically turns around decisions within 24 to 48 hours, though complex or flagged apps can take much longer. This system was built for a world where creating an app required significant technical skill and financial investment, natural constraints that limited submission volume.
Vibe coding obliterates those constraints.
The barrier to creating a submittable app has dropped from thousands of dollars and months of work to essentially nothing. A teenager with a ChatGPT subscription can now generate a weather app, a habit tracker, or a simple game in an afternoon. Multiply that by millions of newly empowered would-be developers worldwide, and the math gets uncomfortable quickly.
Several independent iOS developers told Business Insider that they’ve already noticed review times creeping upward in recent months. Some reported receiving rejection notices that seemed generic or inconsistent, suggesting reviewers may be under pressure to move faster through a growing queue. One developer described getting contradictory feedback on successive submissions of the same app — approved features suddenly flagged as violations, then cleared again on appeal.
Apple declined to comment on whether submission volumes have increased or whether vibe-coded apps are affecting review timelines. The company has historically been tight-lipped about the internal mechanics of App Review, treating it as something between a trade secret and a sacred process.
But the signals are there. In its most recent transparency report, Apple noted it had rejected over 1.7 million app submissions in 2024 for various guideline violations. If AI-generated submissions are growing at even a fraction of the rate that AI coding tool adoption suggests, that rejection number could spike dramatically — and each rejection still requires human review time, appeals processing, and communication with the submitter.
The quality problem compounds the volume problem. Experienced developers understand Apple’s Human Interface Guidelines, accessibility requirements, and privacy rules because they’ve spent years learning them. Someone who asked an AI to “make me a meditation app with a subscription” may have no idea their app violates data collection rules or uses deprecated APIs. The result: more submissions that fail review, more back-and-forth, more strain on a system already operating at scale.
What’s Actually in These Apps — And Why It Matters
The deeper concern among security researchers and veteran developers isn’t just volume. It’s what’s inside the code.
When a human developer writes an app, they generally understand — or at least can explain — every line of code. When an AI generates thousands of lines of Swift or Kotlin in seconds, the person who prompted it often can’t. This creates a troubling accountability gap. If an AI-generated app contains a security vulnerability, collects user data improperly, or includes a hidden dependency that phones home to a third-party server, who’s responsible? The person who typed “build me a fitness app”? The AI company whose model wrote the code? Apple, for approving it?
These aren’t abstract questions. Security researchers have already demonstrated that AI coding tools can introduce subtle vulnerabilities — buffer overflows, improper input validation, insecure data storage — that pass superficial review but could be exploited at scale. A Stanford study published in 2024 found that code generated by large language models contained security flaws at rates comparable to or higher than human-written code, with the added problem that the humans using the AI often believed the code was more secure than it actually was.
Apple’s automated review tools catch some of these issues. But they weren’t designed for a world where millions of people are submitting machine-generated code they don’t understand. The automated scanners look for known patterns — use of private APIs, certain types of data access, obvious malware signatures. They’re less effective at catching novel vulnerabilities or subtle privacy violations buried in AI-generated boilerplate.
Google faces the same challenge with the Play Store, though its more permissive review process means the effects may manifest differently — potentially as a quality crisis rather than a bottleneck. Both companies have invested heavily in automated review capabilities, but neither has publicly addressed how they plan to handle the vibe coding wave specifically.
Some developers have proposed that Apple should require a declaration of AI involvement in app submissions, similar to how the company now requires privacy nutrition labels. Others have suggested a separate review track for AI-generated apps, with additional automated scrutiny and longer review windows. Neither idea has gained official traction.
The indie developer community is particularly anxious. For small studios and solo developers, App Store visibility is already brutally competitive. If the store gets flooded with thousands of low-quality AI-generated clones of popular app categories — and there’s every indication it will — discoverability for legitimate apps could suffer. The App Store’s search and ranking algorithms already struggle with keyword manipulation and review fraud. Adding millions of AI-generated apps to the mix won’t help.
There’s a historical parallel worth considering. In the early days of the App Store, a similar flood of low-effort apps — fart soundboards, flashlight utilities, simple tip calculators — prompted Apple to tighten its review guidelines and raise quality standards. That cleanup took years and alienated some developers in the process. The vibe coding wave could force a similar reckoning, but at a scale orders of magnitude larger.
The Business Model Beneath the Chaos
Follow the money and the picture gets even more complicated. Many vibe-coded apps are being built not as passion projects but as quick monetization plays. The playbook is simple: use AI to generate an app in a popular category, slap on a subscription paywall or ad integration, submit it to the App Store, and hope it catches enough organic traffic to turn a profit before Apple or users notice the quality issues.
This isn’t speculation. Developers on X and in forums like the Indie Dev subreddit have documented a surge in copycat apps that appear to be AI-generated, complete with generic UI patterns, placeholder-quality icons, and subscription prompts that appear before the user has done anything meaningful. Some of these apps are making real money — enough to incentivize more of the same.
Apple takes a 15-to-30 percent cut of all App Store transactions, which creates a perverse incentive. More apps generating subscription revenue means more revenue for Apple, even if the apps themselves are mediocre. But the long-term risk to the platform’s reputation — and to consumer trust in App Store purchases — could outweigh the short-term financial gain. Apple has built its brand partly on the promise that its walled garden keeps out the junk. If that perception erodes, it undermines one of the company’s core arguments against sideloading and alternative app stores.
The timing is particularly awkward. Apple is simultaneously promoting its own AI capabilities through Apple Intelligence and facing regulatory pressure in the EU and elsewhere to open its platform to competing app marketplaces. If the company cracks down too aggressively on AI-generated apps, it risks appearing hostile to the very technology it’s championing. If it doesn’t crack down enough, the quality of its flagship software distribution platform deteriorates.
So what happens next?
The most likely near-term outcome is incremental tightening. Apple will probably update its App Store Review Guidelines to address AI-generated submissions more explicitly, possibly requiring developers to disclose AI involvement and demonstrate understanding of their own code. The company may also invest in more sophisticated automated scanning tools designed specifically to detect common patterns in AI-generated code — the telltale signs of a Claude or GPT-4 output that a trained system could learn to flag.
But these are band-aids on a structural problem. The fundamental tension — between the democratization of app development and the maintenance of quality standards — isn’t going away. If anything, it’ll intensify as AI coding tools get better, faster, and cheaper. The next generation of these tools won’t just write code; they’ll understand Apple’s guidelines, pre-check submissions for compliance, and iterate until the app passes review. At that point, the distinction between a vibe-coded app and a professionally developed one may become genuinely difficult to detect.
For Apple, the vibe coding wave is ultimately a governance challenge disguised as a technology trend. The company built a centralized review system predicated on scarcity — the scarcity of development talent, the scarcity of submission volume relative to review capacity, the scarcity of people willing to invest the time and money to build something worth submitting. AI has made all of those things abundant overnight.
The gates are still standing. But the crowd outside is growing fast.
WebProNews is an iEntry Publication