AI agents – Apple is weighing how to allow AI-agent apps on the App Store without violating review rules that protect users and safeguard revenue.
Apple’s App Store is already crowded with AI-powered apps, but the company is still struggling with a harder question: what happens when AI agents start doing the coding?
Apple is reportedly holding internal discussions about how to incorporate apps that use AI agents into the App Store without breaking the long-standing rules developers have to follow.. The challenge is less about whether AI can be used to build software and more about whether AI can be used to build software inside Apple’s platform.
Artificial intelligence has helped drive a new wave of apps, including those built with agentic coding techniques.. Sources indicate that Apple is comfortable with AI being used to produce apps in general. yet remains cautious when the AI is capable of contributing directly to app creation on iPhone and iPad.
So far, Apple has been blocking a specific category of tools: apps that can be used for “vibe coding,” where AI writes code and effectively generates programs. The reason is tied directly to App Store Review Guidelines, which forbid apps from coding and producing other apps on the device.
At the heart of the issue is the way Apple reviews software before it reaches users.. The review process is designed to safeguard against potential harm, including malware.. If coding apps were allowed to pass review while still being capable of producing new code on-device. it could open a path for malware creation that Apple would not be able to inspect ahead of time.
There’s also an economic and marketplace dimension to the concern.. Some agentic coding apps could create other apps for the user without the need to purchase them from the App Store. potentially undermining the store’s revenue model.. Apple’s dilemma. sources suggest. is that enabling these tools could boost adoption as the apps become more popular. while also putting the broader ecosystem at risk.
In response. Apple is reportedly trying to reconcile its goals with its rules by developing a mechanism intended to bring apps using AI agents into the store while requiring them to meet privacy and security standards.. The proposed direction appears aimed at limiting the most elaborate agentic systems from being available on iPhone.
One example of what Apple wants to avoid is software with broad reach across a user’s system—described in reporting as “OpenClaw-style” capabilities.. A system that enforces stricter privacy and security requirements would. in theory. prevent certain agent designs from gaining the kind of access that could make them difficult to control or evaluate.
It’s also worth noting that Apple’s AI strategy is not confined to App Store rules. With WWDC expected to include substantial AI-related content in June, it remains unclear whether the company’s push will directly involve changes to the App Store or focus on other parts of its ecosystem.
Meanwhile, Apple’s attempt to expand AI options for users appears to run in parallel with a separate effort: giving people more choices in how tasks are handled on iPhone, possibly through the selection of third-party models rather than relying solely on Apple Intelligence.
Rumors indicate that Apple could allow users to choose which third-party models run on their iPhone as an alternative to Apple Intelligence.. According to the reporting. while Apple already integrates ChatGPT so that Apple Intelligence features can run through it—such as Siri queries—more AI models may be supported across the “27-generation” of Apple operating systems.
In this model-selection approach. the selected agents would be able to support functions tied to user workflows. including queries directed to Siri. capabilities in Writing Tools. and generating files through Image Playground.. The broader idea is that Apple would not only provide AI features. but also let users route certain tasks through models Apple approves.
Apple has reportedly been considering this for some time, with earlier rumors dating back to 2024 about an AI App Store. More recent speculation has shifted toward the possibility of an AI section inside the existing store rather than an entirely separate storefront.
The key difference. sources suggest. is that the models Apple allows for deeper access into iOS and macOS would be approved by Apple—and those approved models would face more stringent inspection.. They also would not have the ability to code at all. which directly answers Apple’s current concern about apps being able to generate other apps.
Taken together. the reporting points to a tight balancing act: Apple wants to meet developer demand for agentic AI while protecting the review process that helps prevent malware and guarding the economic structure of the App Store.. The outcome could determine whether AI agents remain mostly assistants that help users with tasks—or whether they ever move closer to building and modifying software in ways Apple can no longer reliably pre-check.
Apple App Store AI agents agentic coding iPhone models privacy and security standards Siri integration malware risk