Apple weighs allowing AI agents in the App Store
The Information reported on May 13, 2026 that Apple is exploring how to admit apps built around AI agents into the App Store without abandoning guidelines that ban apps from generating and executing code on-device, the rule behind Apple's crackdown on so-called "vibe coding" apps (AppleInsider). Apple's concern, per AppleInsider's recap, is that unreviewable generated code could produce malware or let agents build apps that bypass in-app purchases; one proposed constraint is barring any "OpenClaw"-style agent with sweeping system access. This remains single-sourced reporting that Apple has not confirmed, and Apple's WWDC 2026 announcements expanded agent APIs and developer tooling without publicly settling this specific App Store policy question.
The narrower and more useful question here isn't whether Apple will allow AI in the App Store, it already does, it's whether Apple will lift its specific ban on apps that generate and execute code on-device. WWDC 2026 answered an adjacent question (letting third-party apps expose actions to Siri AI, and bringing agentic coding tools into Xcode for developers) but coverage of the keynote does not show Apple settling the original on-device code-execution policy The Information described.
What happened
The Information reported on May 13, 2026 that Apple is discussing how to incorporate apps built around AI agents into the App Store while preserving guidelines that prohibit apps from writing and running code on the device itself, the rule Apple has used to remove "vibe coding" apps (AppleInsider, which recapped and expanded on The Information's reporting). Apple's stated concerns, per AppleInsider, are that such agentic code generation could produce inspectable-free malware and that agent-built apps could let users skip in-app purchases entirely. AppleInsider reports Apple has floated designing a constrained system that would, among other things, bar any "OpenClaw"-style agent with sweeping system-level access. The effort runs alongside a separate, related plan (also from The Information via AppleInsider) to let iOS 27 users choose approved third-party AI models for Siri, Writing Tools, and Image Playground; those approved models reportedly will not have code-execution capability at all.
Technical context
At WWDC 2026 in June, Apple expanded the App Intents framework so third-party apps can expose actions and on-screen context to Siri AI through entity and intent schemas, deprecated SiriKit in favor of App Intents, and brought agentic coding tools built on models from Anthropic, Google, and OpenAI into Xcode 27 for app developers (Apple Newsroom; MacRumors). Those changes govern how apps integrate with Siri and how developers build apps, not whether an end-user-facing app may generate and run code on a device, so they do not appear to resolve the specific App Store Review Guidelines question The Information originally reported.
For practitioners
Teams building apps that embed AI agents should not infer App Store approval from the App Intents and Xcode agentic-coding announcements, since those address a different problem (system-level agent extensibility and developer tooling) than on-device code generation and execution. Until Apple publishes explicit Review Guidelines language, the safest design assumption remains that agent-built code should run server-side or inside a vetted sandbox rather than generate and execute arbitrary code on the device.
What to watch
- •Any explicit update to the App Store Review Guidelines addressing AI-agent or vibe-coding apps.
- •Whether Apple or The Information publishes further reporting confirming or denying the original scoop.
- •The rollout of user-selectable third-party AI models in iOS 27 and whether any are granted code-execution permissions.
- •Enforcement patterns: whether Apple continues removing vibe-coding apps at the same pace reported earlier in 2026.
Key Points
- 1The Information reports Apple is exploring how to admit AI-agent apps to the App Store without lifting its ban on on-device code generation.
- 2Apple's stated worry is that agent-generated code could hide malware or let apps bypass in-app purchases, per AppleInsider's recap of the report.
- 3WWDC 2026 expanded App Intents and Xcode's agentic coding tools but did not appear to resolve this specific App Store policy question.
Scoring Rationale
Downgraded slightly from initial reporting: this remains fundamentally single-sourced (The Information, recapped by many outlets but not independently confirmed or denied by Apple), and more than a month past WWDC 2026, Apple's public announcements addressed adjacent agent-extensibility and developer-tooling questions without visibly resolving the specific App Store policy this story describes. Still relevant to developers given the App Store's scale, but confidence should stay measured given the lack of confirmation or follow-through.
Sources
Primary source and supporting public references used for this report.
View 8 more sources
- Apple Working on Plan to Allow AI Agent Apps on the App Storemacrumors.com
- Apple Prepares App Store for Autonomous AI Agentspymnts.com
- Apple may loosen App Store rules for AI agentsdataconomy.com
- iPhone might be swarming with AI agents sooncultofmac.com
- Apple Explores Ways to Welcome AI Agents in the App Storetheinformation.com
- Apple is working to incorporate AI agents on the App Store, per report9to5mac.com
- Apple may open up the App Store to agentic AIengadget.com
- Apple aids app development with new intelligence frameworks and advanced toolsapple.com
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems