Analysis
Apple released iOS 27, iPadOS 27 and macOS 27 "Golden Gate" on September 14, headlined by a beta version of Siri AI -- Apple's most aggressive overhaul of its voice assistant to date, spanning more than 250 changes across the operating system, multiple outlets reported.
Siri AI ships first as an English-only beta, with French, Japanese, Korean, Portuguese and Spanish support promised for October. The new assistant can hold conversational, context-aware exchanges, pull personal information from across a user's apps, read what's currently on screen to inform its responses, and complete complex multi-step actions spanning multiple apps -- a meaningful jump from the largely command-based Siri that shipped for the past decade. Apple also introduced a standalone Siri app alongside the system-wide assistant.
Hardware support is notably narrow: the new Siri AI only runs on iPhone 15 Pro and newer, meaning just three and a half generations of iPhones -- roughly the newest quarter of Apple's active installed base -- can access it at launch. iPad support is even tighter, limited to the latest iPad mini and iPad Air or Pro models running an M1 chip or newer. That's a deliberate compute-gating decision: running a genuinely conversational, context-aware assistant on-device or via Apple's private cloud compute requires silicon Apple didn't ship broadly until relatively recently.
“iPad support is even tighter, limited to the latest iPad mini and iPad Air or Pro models running an M1 chip or newer.”
The gap between Apple's Siri ambitions and shipped reality has been one of the most closely watched storylines in consumer AI for two straight product cycles -- Apple originally previewed a more personalized, context-aware Siri years earlier and then delayed it repeatedly, drawing direct comparisons to how quickly OpenAI's ChatGPT and Google's Gemini assistant shipped comparable conversational capabilities. This release is Apple's clearest answer yet to that criticism, even if the English-only, newer-hardware-only rollout means most of Apple's installed base still won't experience it at launch.
Third-party developers are already building on top of Siri AI's new capabilities: fashion app Daydream, covered elsewhere in this issue, shipped a Siri-powered natural-language shopping search feature the same day iOS 27 became available -- an early signal that Apple's platform-level AI investment is opening real third-party integration surface rather than staying locked to Apple's own apps.
Apple faces direct competitive pressure from Google, whose Gemini has been built into Android as a system-level assistant with broader multi-language support from an earlier stage, and from OpenAI's ChatGPT, which many iPhone users already run as a standalone app for tasks Siri still can't reliably handle. Whether Siri AI closes that capability gap depends heavily on real-world reliability once it's in the hands of millions of users, not on the feature list Apple previewed at launch.
A beta label matters here: Apple explicitly positioned Siri AI as unfinished, English-only, and hardware-restricted, which means the headline "Apple ships conversational Siri" overstates how many iPhone users can actually use it today. The real rollout -- broader language support, older-hardware compatibility if it ever comes, and reliability at scale -- happens over the following months, not on release day.
How Siri AI performs against ChatGPT and Gemini in real independent usage over the next few weeks, and whether Apple hits its own October timeline for the five additional languages, are the concrete tests of whether this closes the gap Apple has been criticized over for two product cycles.
