Apple may let users choose an AI model, but still keep the keys to the iPhone

Apple's reported iOS 27 plan is not a simple embrace of openness. The company is exploring third-party chatbot extensions inside Apple Intelligence, which would let users choose models beyond Apple's own stack for some assistant functions.

That sounds like a win for users, but the important detail is the gate. A system-wide AI extension still has to pass through Apple's privacy rules, permission surfaces and platform APIs. Apple is not handing over the operating system; it is designing a controlled window.

The move also shifts risk. If a third-party model produces a weak answer, unsafe recommendation or privacy dispute, Apple can frame the issue as provider-specific rather than a failure of Apple Intelligence itself.

For developers, the interesting part is distribution. Being available inside iOS gives a model provider a path to mainstream users without asking them to install yet another app. For users, the harder part will be knowing which model to trust for which task.

If Apple ships this as expected, the signal is clear: even the most closed consumer platform now sees AI choice as strategically useful. But choice inside a curated gate is still gatekeeping.

For source context, compare The Verge, NIST AI RMF and OECD AI Principles.