Apple is discussing changes to App Store rules that would allow publishing applications with so-called agentic AI — artificial intelligence that independently performs actions in other programs on behalf of the user. This is reported by The Information, citing sources familiar with the matter inside the company.
The difference between a regular chatbot and an agent is fundamental. A chatbot answers questions. An agent books tickets, sends emails, moves files, fills out forms. The user sets a task, and then the system acts on its own. There is no need to confirm each step.
This is where the real tension of this decision emerges: the App Store has been built for decades on the logic of "every action goes through a human." App review, permissions, system restrictions — all of this has restrained automated actions by third-party software. An AI agent by definition breaks this logic.
The question is not whether it is convenient to delegate routine work to a machine — it obviously is. The question is what limits Apple intends to set for agents that will have access to email, calendar, banking applications, and geolocation simultaneously.
So far, the company has not disclosed either the technical requirements for such applications or the mechanism for auditing an agent's actions after the fact. All that is known is that discussions are ongoing.
Competitors are not waiting. Google is already integrating agentic capabilities into Android through Gemini, Microsoft is promoting Copilot as an agent in Windows and Microsoft 365. Apple, if it does not open up its platform, risks losing developers who will go where there are fewer restrictions.
The key question that remains unanswered so far: will the user get a tool to review what the agent did on their behalf — and will this tool be a mandatory requirement for acceptance into the App Store?
