Siri AI: What’s Confirmed, What’s Restricted, and What It Means in Practice

Siri AI is Apple’s fully rebuilt virtual assistant, powered by the next generation of Apple Intelligence. It launched in beta with the iOS 27 / macOS 27 family in September 2026 and has since expanded language support. It is not a minor upgrade to the old Siri; it is a new system with personal-context understanding, onscreen awareness, system-wide app actions, and a dedicated conversation app that syncs privately via iCloud.

This piece sticks to verified information from Apple’s official channels and documented regulatory outcomes. It avoids rumor and focuses on what is actually shipping, which devices support it, where it is blocked, and the practical implications for anyone evaluating on-device or hybrid assistants.

Core Capabilities That Are Live

According to Apple’s documentation and support pages:

• Personal context: Siri AI can search across Messages, Mail, Photos, Notes, Calendar, and other apps to surface information you already have (e.g., a hotel confirmation buried in email or photos from a specific trip).

• Onscreen awareness: It can answer questions or take actions related to what is currently visible on the device.

• App actions: It can draft emails, edit and share photos, create reminders or events, and perform multi-step tasks across apps.

• World knowledge: It can pull current information from the web when needed and continue the conversation with follow-ups.

• Dedicated Siri app: Conversation history syncs privately across supported devices via iCloud so you can start on one device and continue on another.

• Writing tools: Users can generate or refine text almost anywhere they type; the system can adapt tone and style based on prior communication patterns with a recipient.

• Visual Intelligence: Expanded to more devices and interaction methods (Camera mode on iPhone, screenshot on iPad, keyboard shortcut on Mac, gaze + speech on Vision Pro).

These features run on Apple’s third-generation Foundation Models. Basic (cloud-assisted) processing is available on a broad set of recent devices; a higher-capability advanced on-device model (more expressive voices, improved dictation) requires newer silicon and higher memory thresholds (roughly A19 Pro / M3 or later class hardware with sufficient unified memory).

Device Support Snapshot

Supported platforms include recent iPhones (15 Pro and later generations, including newer “Air” and “Duo” models), iPads with M1 or later (plus the A17 Pro mini), Macs with M1 or later (plus the A18 Pro MacBook Neo), Apple Watch Series 9 / Ultra 2 / SE 3 when paired with a compatible iPhone, and Apple Vision Pro. Exact cut-offs for the advanced on-device model are stricter and documented on Apple’s support pages.

Not every feature is identical across the fleet. Expressive voice customization and the strongest on-device dictation improvements are limited to the advanced model tier.

Regional Reality: EU and Mainland China

Two major markets remain restricted:

• European Union: Siri AI is unavailable on iOS, iPadOS, and watchOS. Apple attributes this to Digital Markets Act (DMA) interoperability requirements. Apple states that complying while preserving its privacy and security model would require giving third-party assistants system-level access it considers unacceptable. EU regulators have publicly rejected Apple’s characterization, stating the decision not to ship is Apple’s alone and that nothing in the DMA prevents a compliant launch. macOS and visionOS users in the EU can access Siri AI when language settings are supported. No firm timeline has been given for iPhone/iPad availability.

• Mainland China: Full Siri AI is not available under the global configuration. Apple has instead begun rolling out Apple Intelligence features in China via a partnership that routes certain requests (including knowledge and document/image understanding) through Alibaba’s Qwen models. This is a region-specific architecture, not a global Siri AI deployment.

Language support has expanded beyond English. As of late October 2026 updates, French, Japanese, Korean, Portuguese, and Spanish are available in supported regions (with further languages planned). Language and Apple Account region both matter; setting a supported language is not always sufficient if the account region is restricted.

Architecture Notes Relevant to Practitioners

Apple emphasizes a hybrid design: on-device models where possible, Private Cloud Compute for heavier requests, and strict claims that personal data processed in the cloud is not stored or accessible to Apple. The system orchestrator and Spotlight index keep many actions local. Third-party developers can now integrate more deeply (Developer Preview opened more broadly in early November 2026 betas), including richer entity surfacing and app actions.

Earlier reporting established that Google’s Gemini technology formed the technical foundation for the next-generation Apple Foundation Models used in this release. Apple continues to position the resulting models as its own, running under its privacy architecture. In China the picture is different: Qwen is the explicit partner for the knowledge and multimodal pieces that are live there.

Usage limits apply to certain cloud-backed features; Apple has indicated expanded paid access will arrive later.

Practical Takeaways

If you are evaluating Siri AI for real workflows:

1. Confirm your exact hardware and the language/region of the Apple Account. The feature matrix is not uniform.

2. Treat the EU restriction as structural for now. Do not plan iPhone/iPad deployments around Siri AI availability in the bloc until Apple and regulators reach an agreement.

3. In China the relevant capability is the Qwen-backed Apple Intelligence path, not global Siri AI. Teams operating in both regions will need dual evaluation paths.

4. For developers, the expanded integration surface (Spotlight entities, app actions, system-wide context) is the more durable change. Test against the current betas rather than marketing screenshots.

5. Privacy posture remains a core differentiator Apple markets. Independent verification of Private Cloud Compute claims continues to be possible in principle; treat that as an ongoing operational check rather than a one-time claim.

Siri AI is a significant step for Apple’s assistant stack, but its availability map is still fragmented by hardware tier and regulation. The cleanest way to assess it is the same approach we use for any local or hybrid model: verify the exact model tier on the target device, measure the tasks that matter to your workload, and separate marketing language from the documented support matrix.