Apple's New Siri Runs on Google Gemini — And Feels Late
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Apple launched Siri AI in the iOS 27 consumer beta in July after multiple rounds of delays, with the public rollout expected in September alongside the official iOS 27 release
- The new Siri supports natural back-and-forth conversations with adjustable voice expressivity and pacing, accepts typed input, and works through a dedicated app that references past conversations
- Siri can locate on-device content — photos, emails, contacts, texts, calendar appointments — even when users lack specific details, such as pulling up the last receipt saved on the iPhone without naming its source or subject
- Apple partnered with Google to use Gemini AI models to train its proprietary Apple Foundation Models, which are designed to run on Apple silicon and its private cloud compute infrastructure
- Despite the functionality, Siri's arrival feels anticlimactic because the AI race has surged ahead — AI tools are now coding software, completing multistep tasks, and operating computers autonomously
- Siri can be used inside the camera viewfinder to identify real-world objects and split restaurant bills with friends, and now consistently plays the requested song, podcast, or audiobook on the user's preferred app
Why it matters: Apple's dependence on Google's Gemini to train its foundation models is a significant concession — the company known for in-house silicon and software is now borrowing core AI from a primary competitor. Users gain a finally-functional assistant, but Apple entered an AI conversation that has already evolved past chatbots into agentic systems that code and operate computers, meaning Siri's delivery fulfills an outdated promise.
