Siri AI is becoming more capable of controlling the iPhone, but being able to follow complicated instructions is not quite the same as working independently. That distinction is increasingly important as Apple considers how its assistant should evolve alongside autonomous AI agents.
In a recent Power On discussion, Bloomberg’s Mark Gurman described the expanded Siri controls arriving with iOS 27.3 as request-based rather than truly autonomous. His November 8 newsletter explores the possibility of a more agent-like Siri as Apple looks ahead to iOS 28.
That may sound like a small distinction, but it changes what customers should expect. Asking Siri to find a message and move it into a folder is a defined instruction. Asking an AI agent to organize an inbox, identify important messages and keep it tidy over time involves a continuing goal, multiple decisions and potentially access to several services.
What separates Siri AI from an autonomous agent?
Apple’s latest iOS 27.3 improvements allow more detailed actions inside apps, including support for developers to define sequences of tasks. This makes the assistant more useful for everyday work, but the user remains responsible for initiating those interactions.
By comparison, a system designed around ongoing tasks could periodically check for changes, make progress between conversations and return with results. That could be useful for travel planning, comparing options or keeping track of an unfinished project. It also brings greater risk if the assistant misunderstands a request or takes an action the user did not intend.
Privacy could decide whether Apple’s approach succeeds
Apple’s strongest potential advantage is the amount of personal context already available on its devices, together with its established emphasis on privacy and device-level protections. Yet useful autonomy may also require access to outside websites, connected accounts and services that Apple does not control. Secure processing alone cannot settle questions about permission, mistakes or accountability.
A practical agent would need transparent controls showing what it can access, which actions it is allowed to take and when the owner must approve a decision. The more useful it becomes, the more important those safeguards are likely to be. Apple also needs to avoid making routine tasks more complicated for customers who simply want quick answers from Siri.
The immediate priority is for Apple’s new app controls to work reliably. If those foundations prove dependable, a future autonomous Siri could offer something meaningfully different from today’s assistant. Until Apple details such a product, however, any iOS 28 agent features remain possibilities rather than promises.
