Apple WWDC 2026: Siri Gets Rebuilt With Google Gemini
Apple's voice assistant got a new name, a new app, and a Google engine under the hood
Apple unveiled a completely rebuilt voice assistant at WWDC 2026 on June 8. Siri AI - the company's new name for the product - runs on Google Gemini models and arrives with a dedicated standalone app for the first time in Siri's history. Alongside iOS 27, iPadOS 27, and macOS 27, Apple rolled out AI updates across Photos, Safari, Passwords, Messages, and Shortcuts. Developer testing starts today.
Apple built the next generation of its foundation models in collaboration with Google, licensing Gemini's architecture for what reports put at around $1 billion per year. Requests requiring more compute route to Private Cloud Compute servers instead of the device chip. Private Cloud Compute stores no personal data from those sessions. Outside researchers can inspect that privacy promise at any time.
Siri AI Can Now Take Action Across Apps
Old Siri answered questions. New Siri AI takes action, holds back-and-forth conversations, and accepts typed input for the first time. Ask it to surface a flight confirmation number from your email during a call and the Phone app shows it automatically via Call Context. Tell Messages to find photos from a specific trip and Siri searches the entire photo library, recognizing keywords, locations, and people.
Claude arrives on the iPhone through Apple's new multi-AI Extensions framework. iOS 27 developer betas include a settings panel and App Store section letting users swap between ChatGPT, Claude, and Google Gemini inside Siri. Apple has held direct talks with Anthropic, OpenAI, and Google about granting access to the framework. Full entitlements are not confirmed yet. For developers, that is the part worth watching - an official, Apple-sanctioned routing layer for competing AI models built into the OS itself.
Photos, Safari, and Passwords All Get AI Features
Spatial Reframing lets users drag a photo to shift perspective after the fact, generating new pixels where the perspective changed while keeping the rest of the scene consistent with the original shot. Extend fills missing edges. Safari organizes open tabs into topics automatically and adds Notify Me alerts for price drops or page changes. Passwords navigates to websites on a user's behalf and resets weak or compromised credentials with one tap.
Photorealistic image generation arrives in Image Playground, running on Private Cloud Compute servers rather than the device. Apple embeds a hidden SynthID watermark in every AI-generated or AI-edited photo, including edits made inside the native Photos app. Shortcuts gains a Describe a Shortcut feature that writes automation steps from a plain-text description. Genmoji quality improves too.
Who Gets It and When
Apple Intelligence in iOS 27 requires iPhone 16 or later, iPhone 15 Pro, iPhone 15 Pro Max, any iPad with M1 or later, or any Mac with M1 or newer. A public beta opens next month. Users get the full release this fall in English first, with more languages to follow.
Siri AI will not launch in China. In the EU, Apple is delaying Siri AI for iOS and iPadOS due to DMA regulations, though Mac, Apple Watch, and Apple Vision Pro users in Europe keep access. Features relying on heavy cloud compute, including photorealistic image generation, carry daily usage limits. Higher limits come with iCloud+ plans.
Apple's partnership with Google puts Gemini at the heart of the world's most-used mobile platform. No other AI lab got that contract. ChatGPT hit 1 billion monthly users in May 2026 while simultaneously losing market share, and the standings among the major labs are shifting faster than most predicted. For where OpenAI, Anthropic, and Google each stand today, read our OpenAI vs Anthropic vs Google DeepMind breakdown.
