Apple weighs allowing AI agents in the App Store

The Information reports that Apple is exploring ways to welcome apps that embed AI agents into the App Store, according to a summary in AppleInsider. AppleInsider reports The Information says Apple has been blocking apps that provide functionality for writing code or creating programs, citing the App Store Review Guidelines' prohibitions on on-device code generation and execution. AppleInsider adds that the company worries such agentic coding features could be used to create malware or to generate apps that sidestep App Store purchases, and that Apple is discussing technical and policy mechanisms to limit those risks.

Editorial analysis - technical context: Agentic apps that generate or execute code on a device substantially expand the attack surface compared with static mobile apps. Industry-pattern observations note that controlling runtime behavior and sandbox escapes is hard on mobile platforms, and that containment typically requires a mix of static review, runtime monitoring, and strict sandboxing or server-side execution.

The decision matters because the App Store is a primary distribution and monetization channel for mobile software. Allowing agentic AI within the App Store under constrained conditions could enable new developer workflows and consumer features while exposing Apple to new moderation, security, and revenue-protection tradeoffs. Observers have previously flagged that platform owners balance enabling innovation with enforcing ecosystem rules; reporting frames Apple's internal discussions as part of that broader tension.

Observers will watch WWDC in June for any Apple announcements related to AI tooling or policy; AppleInsider notes WWDC may include AI-related content. Additional signals to follow include any formal changes to App Store Review Guidelines, new developer-facing APIs or sandboxing controls, and whether Apple publishes explicit rules for on-device code generation or requires server-side execution for agentic features.

Editorial analysis: For practitioners, the practical implications will center on where compute and execution are permitted. If platform policy forces model execution server-side, that affects latency, cost, and data flows. If Apple instead creates vetted execution sandboxes, developers will need to design agentic features to satisfy review and runtime constraints.

This summary relies on reporting in The Information as recapped by AppleInsider.