Apple unveils Siri AI overhaul at WWDC

Apple unveiled a major overhaul of its voice assistant at its Worldwide Developers Conference on June 8, 2026, introducing a product it brands "Siri AI," per Reuters and Apple's product page. The update is part of Apple Intelligence and promises more conversational, back-and-forth interactions, the ability to act inside apps, on-screen awareness, and visual features such as photo editing and Visual Intelligence across Photos, Messages and Safari, alongside a standalone Siri app. Mike Rockwell, the Apple executive who leads Siri, said Siri is "now a profoundly more capable assistant that helps you find what you need and gets more done," per WWDC coverage. The Week reports the update reaches developers immediately, with a public beta next month and a full customer launch in the fall; Apple's page states Siri AI is "coming later this year" in English.

Apple said Siri AI's cloud intelligence is built in collaboration with Google, extending a deal first reported in January 2026 and described in reporting as worth about $1 billion a year; Apple has not confirmed the commercial terms. The exact dependency is contested: the original framing was that Apple's cloud models are "based on" Google's Gemini, while June 2026 reporting (for example AppleInsider) describes them as custom Apple Foundation Models distilled from, or refined using outputs of, Gemini rather than running Gemini at runtime. On-device requests run on Apple's own models, while heavier requests route to Apple's Private Cloud Compute, with the top server model reportedly running on NVIDIA GPUs in Google's cloud. Either way, the arrangement is notable because it shows a leading platform vendor leaning on a rival's frontier-model technology rather than relying solely on in-house systems.

Reuters and other outlets frame the launch as Apple trying to close a capability gap with chat-first assistants such as OpenAI's ChatGPT and Google's Gemini, after the company's earlier Siri upgrades slipped. For engineering and product teams, a hybrid design that mixes a third-party-influenced cloud model with on-device context affects latency, data-residency, debugging and change-management decisions: external dependencies can accelerate feature delivery but tie roadmaps and costs to another vendor.

Vendors commonly pair cloud models for broad knowledge with device-side signals for personalization while emphasizing privacy-preserving architectures. Apple stresses on-device processing and Private Cloud Compute in its materials, but independent verification of data flows typically becomes possible only once betas and documentation ship.

Reuters reports Siri AI will not be available on iOS and iPadOS in the EU at launch, and will not be available in China initially, leaving open how regulation, including the EU's Digital Markets Act and China's model-approval regime, shapes availability and data handling.

• •Developer surfaces: new APIs and SDKs for the standalone Siri app and app-action frameworks announced on Apple's product page.
• •Model behavior: latency, hallucination rates and auditability of the cloud path once public betas open.
• •Data governance: Apple's documentation on what runs on-device versus in Private Cloud Compute, and any transparency reporting.
• •Regional expansion: whether and when Apple enables Siri AI in the EU and China.