Ask Siri what a relative suggested for an upcoming visit and it can now answer from your own inbox. That kind of personal recall sits at the middle of the Siri AI that arrived with iOS 27 on September 14, ending a wait that ran past two years and several delays.
Underneath, Apple turned to Google: the rebuilt assistant runs on Gemini models rather than the older stack and its limited ChatGPT handoff. Beta access starts in English, with French, Japanese, Korean, Portuguese and Spanish due in October.
It reads context from mail, Messages and Photos, and takes action inside apps – drafting in Outlook, finding a homework assignment in Notability, filing a restaurant note in Tripsy. It also works on whatever is visible on screen, so a sports schedule can become calendar entries. Away from the wake phrase, the side button and a swipe from the Dynamic Island both open a session.
The same intelligence spread across Apple’s hardware. Spotlight and context menus carry it on Mac and iPad, Vision Pro renders it as a 3D object to look at, and a Siri app syncs conversation history through iCloud. Photos gained spatial reframing, Writing Tools drafts from a description, and Visual Intelligence shipped beyond the iPhone. Reviewers who had written Siri off say they are using it again for multi-step requests. Generated images will carry SynthID markers in a later release, and users under 13 are excluded.
