Apple Launches Revamped Siri Powered by Google's Gemini, Excludes EU and China at Launch

Apple has released a beta version of its long-delayed Siri overhaul, built on Google's Gemini models, as part of the iOS 27, iPadOS 27, macOS 27, watchOS 27, and visionOS 27 software releases. The assistant, referred to by Apple as "Siri AI," launches in English only, with support for French, Japanese, Korean, Portuguese, and Spanish planned for next month.

The new Siri runs partly on-device and partly through Apple's Private Cloud Compute infrastructure. According to Apple, personal data processed through this pipeline is neither stored nor made accessible to the company. The underlying foundation model shift — from Apple's own systems to Google's Gemini — marks a significant change in strategy after years of internal delays that reportedly cost employees their jobs.

What the new Siri does

Siri AI is designed to understand personal context pulled from messages, emails, and photos, read on-screen content, and retrieve information from the web. Through system-wide app actions, it can draft emails, edit photos, and move information between apps — Apple's example involves finding a recipe mentioned in Messages and transferring the ingredient list into Reminders.

The assistant also integrates with third-party apps including WhatsApp and Audible, with support for Outlook, Notability, and Tripsy coming soon. Users can invoke Siri via "Hey Siri," the side button, Dynamic Island, or Spotlight on iPad and Mac. A dedicated Siri app now syncs conversation history across devices via iCloud. Apple has also added new voice options and an upgraded dictation feature powered by its on-device model, AFM Core Advanced.

Beyond Siri, Apple Intelligence features are expanding elsewhere in the OS: Photos gains Spatial Reframing, Extend, and a reworked Clean Up tool; Image Playground now generates photorealistic images via Private Cloud Compute and will support the SynthID watermarking standard; Safari adds topic-based tab sorting; and Shortcuts can be built using plain-language descriptions. Later this year, Apple plans Audio Intelligence features for Apple Watch Series 12 and Ultra 4, along with a redesigned Health app.

Early reactions are mixed

TechCrunch's Ivan Mehta, who tested iOS 27 betas, called the new Siri a genuine step forward for multi-step requests and screen-context awareness, though he noted bugs like notes being filed incorrectly. Six Colors' Dan Moren reported practical wins — such as Siri locating an old podcast episode via CarPlay based on topic alone — but flagged ongoing misunderstandings, occasional hallucinations, and an assistant that pushes writing suggestions too aggressively. MacStories' Federico Viticci praised Siri's screen-reading capability but said it repeatedly failed to answer questions about personal context despite the relevant information being present in his messages, with some responses based on outdated data.

No launch in the EU or China

Apple confirmed that Siri AI and all features built on it will not be available in the EU across iOS, iPadOS, or watchOS at launch. The company says it is working on an approach that satisfies privacy and security requirements in the region. The feature is also excluded from China while regulatory approval remains pending. Apple noted that certain server-side features carry daily usage caps, with expanded access expected to become a paid tier in the future.

What this means

Apple's decision to build its flagship assistant on a competitor's foundation model underscores how far behind its in-house AI effort had fallen. The EU exclusion signals continued friction between Apple's privacy architecture and European regulatory demands — likely tied to the Digital Markets Act's interoperability requirements around Private Cloud Compute and third-party model access. For now, EU and Chinese iPhone users get the old Siri while the rest of the world tests whether Gemini can finally make Apple's assistant competent.