An internal Apple team is building a system to support third-party AI agents in the App Store while enforcing the company’s privacy and security standards, according to a report from The Information on Tuesday. The effort signals Apple’s first structured response to the agentic AI wave that has reshaped software distribution over the past year.
The Problem Apple Is Solving
AI agents create a specific tension with Apple’s review model. Traditional apps submit code, Apple reviews it, and the approved binary ships. Agents break that loop. According to The Information, some agents can “spin up smaller apps on the spot to perform a wide variety of tasks,” generating new functionality after Apple has already approved the parent app. That violates existing App Store guidelines prohibiting apps from executing code that alters their own behavior or the behavior of other apps.
Apple already enforced this boundary in March when it blocked updates for popular vibe coding apps that let users build apps and websites through natural language prompts, as MacRumors reported. The new framework is a more systematic answer: rather than blocking agent-powered apps outright, Apple is designing guardrails that allow agent behavior within defined limits.
OpenClaw as Cautionary Reference
The report specifically cites OpenClaw as an example of what Apple wants to avoid. People briefed on the matter told The Information that Apple’s staffers are designing the system to “prevent the more freewheeling behavior some users of agentic systems such as OpenClaw have experienced, where agents can go haywire and delete all of a user’s emails.”
That framing positions Apple’s approach as the governed alternative to OpenClaw’s open architecture. Where OpenClaw lets agents operate with broad system access and minimal gatekeeping, Apple’s framework would enforce platform-level constraints on what agents can do, when, and with what permissions.
Siri Overhaul and Developer Outreach
The agent framework is part of a broader AI push. MacRumors reports that Apple has started contacting app developers to integrate capabilities like booking flights and sending calendar invites into a rebuilt version of Siri for iOS 27. Apple plans to let users select from multiple chatbot backends, not just OpenAI’s ChatGPT, with models from Anthropic and Google as potential options for Image Playground and Writing Tools.
Some developers are hesitant to participate. According to MacRumors, Apple is telling developers it won’t charge commissions during the early stages of the Siri integration partnership, but fees remain a possibility later. Baidu, Alibaba, and Tencent have all held talks with Apple about Siri integration in China but are wary of eventual commission exposure.
Revenue Context
The financial incentive is significant. Generative AI apps generated nearly $900 million in App Store fees for Apple in 2025, according to MacDailyNews, with projections exceeding $1 billion in 2026 from in-app subscriptions alone. An agent-friendly App Store would expand that addressable market substantially, capturing revenue from autonomous workflows that currently bypass the App Store entirely.
The Platform Governance Question
Apple’s timing matters. The agent ecosystem has already bifurcated between OpenClaw’s permissive, local-first model and enterprise-controlled platforms from Anthropic and Microsoft. Apple entering with a governed App Store framework creates a third lane: platform-mediated agents where Apple controls the review, distribution, and permission boundaries.
Whether developers will accept those constraints depends on what Apple actually ships. A WWDC announcement is possible when the conference opens June 8, according to 9to5Mac and The Information, though the company may not be ready to unveil the full system by then.