Apple has fundamentally rebuilt Siri by partnering with Google to co-develop the AI models powering the new "Siri AI" assistant, announced at WWDC26 on June 8, 2026. The move represents a significant departure from Apple's traditional approach of controlling all its technology in-house. Siri AI now draws on foundation models created in collaboration with Google's Gemini family, adapted to run both on Apple devices and through Apple's Private Cloud Compute servers.

Why Did Apple Partner With Google for Siri's Brain?

For over a decade, Apple has marketed itself as a company that controls every layer of its technology stack, from silicon to software. But when it came to rebuilding Siri into a genuinely useful AI assistant, Apple decided it needed Google's expertise. According to reports, the companies entered into a multi-year partnership valued at roughly $1 billion annually to create custom Gemini models tailored for Siri.

The collaboration goes deeper than just licensing a model. Apple and Google jointly developed what Apple calls "Apple Foundation Models," which are custom-built versions of Gemini technology optimized to run efficiently on iPhones, iPads, Macs, and Apple Vision Pro devices. Craig Federighi, Apple's senior vice president of software engineering, stated plainly on stage: "This year we embarked on a deep collaboration with Google, leveraging the technologies behind their Gemini family of models. Together, we created the next generation of Apple Foundation Models".

Craig Federighi, Apple's senior vice president of software engineering

What Can the New Siri AI Actually Do?

The rebuilt Siri represents a complete teardown and reconstruction from the ground up. Unlike the previous version, Siri AI can now understand personal context by searching across your messages, emails, and photos. It can answer questions on nearly any topic using live web knowledge, and it can see what's on your screen to take action on your behalf.

Apple demonstrated practical examples of Siri's new capabilities:

• Email Search: Find a hotel confirmation number buried in an old email thread from months ago
• Photo Recognition: Pull up photos from last weekend by identifying who appears in them
• Visual Intelligence: Point your iPhone camera at a restaurant and ask what it is, then split the bill with friends using Apple Cash
• Writing Assistance: Draft an email from a one-line description or rewrite it to match your usual tone with a specific contact
• Onscreen Awareness: See what's displayed on your screen and act on it, such as calling an airline and automatically providing your booking reference

The new Siri also gets a dedicated app that syncs your conversation history privately across devices through iCloud, allowing you to start a conversation on your Mac and continue it on your iPhone. You can invoke it the traditional way by saying "Hey Siri" or pressing the side button, or by swiping down from the Dynamic Island on newer iPhones.

On higher-end devices like the iPhone 17 Pro and Pro Max, iPad with M4 chip, Mac with M3 and at least 12GB of memory, and Apple Vision Pro with M5, users get additional capabilities including more expressive and adjustable voices, improved dictation accuracy, and stronger natural language understanding.

How Does Apple Keep Your Data Private With Google's Technology?

This is where the tension emerges. Apple has spent years positioning itself as the privacy-conscious alternative to Google, whose business model relies on advertising powered by user data. Now Apple is building its most important AI assistant on technology co-created with Google. Apple's defense centers on its architecture and infrastructure.

Siri AI operates on a hybrid model: some processing happens directly on your device, while other tasks run through Apple's Private Cloud Compute infrastructure. Apple says this approach ensures that personal data is only used to execute your immediate request and is never stored on Apple's servers or accessible to third parties. The company has invited outside experts to verify these privacy guarantees at any time.

A new "system orchestrator" sits at the center of the architecture, deciding what runs locally on your device versus what gets processed in the cloud. Apple's Spotlight index and what the company calls the App Toolbox remain entirely on device.

However, there's one Google fingerprint that extends beyond the model itself. Every photo you edit or generate with Apple Intelligence now receives a hidden SynthID watermark, which is Google DeepMind's provenance technology embedded directly into the Photos and Image Playground apps. This watermark helps track the origin of AI-generated images but represents another layer of Google technology woven into Apple's system.

What Are the Practical Improvements to Apple Intelligence Beyond Siri?

The new architecture brings upgrades across Apple's entire ecosystem. Photos now includes genuinely advanced editing capabilities: Spatial Reframing lets you shift the camera angle after you've taken the shot, Extend fills in a cramped frame, and Clean Up removes photobombers more effectively than before. Image Playground can finally produce photorealistic images, running on Private Cloud Compute.

Safari gains the ability to organize runaway tabs into topics, watch a web page and notify you when a price drops or a stock returns, and even build a custom browser extension from a plain-English description. The Passwords app now fixes weak or breached logins automatically, with Apple Intelligence and Safari quietly signing in and upgrading accounts on your behalf, which Federighi described as happening "agentically".

Under the hood, Apple reports significant performance improvements: iPhone and iPad apps launch up to 30% faster, photos load up to 70% faster after capture, and AirDrop transfers move up to 80% quicker.

How Are Developers Getting Access to These New AI Capabilities?

Apple is opening up its new foundation models to developers through updated frameworks and tools. The Foundation Models framework now serves as a single native Swift API that supports more powerful on-device models with image input, server models, and the ability to build custom skills.

Developers enrolled in the App Store Small Business Program with fewer than 2 million total first-time downloads can access the next generation of Apple Foundation Models running on Private Cloud Compute at no cloud API cost. Developers can also choose to integrate models from other providers like Claude and Gemini, or any provider that implements the new language model protocol.

A brand-new framework called Core AI is designed specifically for running custom models on device. Core AI provides an architecture optimized for the unified memory and Neural Engine of Apple silicon, allowing developers to deploy full-scale large language models locally without relying on cloud servers.

Xcode 27, Apple's development environment, takes a major step forward with agentic coding capabilities. The new version brings the full power of models and agents from Anthropic, Google, and OpenAI directly into a developer's workflow. Conversations with coding agents feature interactive planning, multiturn question-and-answer exchanges, and a canvas that renders code changes and previews side-by-side. Coding agents can now validate their own work, allowing them to run autonomously for longer periods by writing and running tests, trying ideas in isolation with Playgrounds, and checking visual changes with previews.

Steps to Understand Apple's New Developer Tools

• Foundation Models Framework: Developers can now integrate Apple's next-generation foundation models or choose from Claude, Gemini, or any provider implementing the new language model protocol, giving flexibility in choosing AI backends
• Core AI Framework: A new framework optimized for Apple silicon's unified memory and Neural Engine, enabling developers to deploy full-scale large language models directly on devices without cloud dependency
• Xcode 27 Agentic Coding: Developers can work with AI agents from Anthropic, Google, and OpenAI in an integrated environment that supports interactive planning, code validation, and autonomous task execution
• Dynamic Profiles: New capabilities allow developers to update how models interact with their apps on the fly, enabling more adaptive AI experiences without requiring app updates
• Small Business Program Benefits: Developers with fewer than 2 million first-time App Store downloads can access Apple Foundation Models on Private Cloud Compute at no cost

Susan Prescott, Apple's vice president of Worldwide Developer Relations, explained the company's philosophy: "Developers are at the heart of the Apple ecosystem, and our goal is to provide them with the best possible tools and technologies to build the future. With new intelligence frameworks and agentic coding in Xcode 27, developers have the tools they need to focus on what they do best: bringing their incredible ideas to life".

Susan Prescott, Apple's vice president of Worldwide Developer Relations

What's the Bigger Picture Here?

Apple's decision to partner with Google on Siri AI marks a watershed moment for the company. For years, Apple has differentiated itself by controlling its entire technology stack and emphasizing privacy as a core value. The partnership reveals that even Apple recognizes it cannot compete in cutting-edge AI without collaborating with the companies that have invested the most in large language models.

The move also highlights a broader industry trend: the most capable AI systems today require massive computational resources and training data that only a handful of companies possess. Apple's solution was not to build everything from scratch, but to partner with Google and adapt the resulting models to run on Apple's privacy-focused infrastructure.

Whether this hybrid approach truly delivers on Apple's privacy promises remains to be seen. The company has invited outside experts to verify its claims, and that scrutiny will likely intensify as Siri AI rolls out to hundreds of millions of devices later this year.