Apple outlines next generation of Apple Intelligence, expanding AI into Photos, Siri, Safari and more

Apple on Monday introduced the next generation of Apple Intelligence, positioning it as an AI layer that is more tightly integrated into everyday apps across iPhone, iPad, Mac, Apple Watch, AirPods and Apple Vision Pro. In a press release, Apple said the update is powered by a new architecture that places its latest “Apple Foundation Models” deep inside the company’s platforms, while also emphasizing privacy protections designed to keep personal data from being stored or shared with Apple when cloud processing is used. The company also said the feature set will be available for developer testing starting immediately, with general user availability targeted for this fall.

The release highlights new capabilities in Photos, Safari and Passwords, alongside an expanded image generation tool called Image Playground and an upgraded Siri. Apple’s Craig Federighi, senior vice president of Software Engineering, framed the effort as a continuation of Apple’s goal to make advanced technology “helpful and intuitive,” describing Apple Intelligence as centered on user needs, grounded in personal context, and built with privacy at every step.

In Photos, Apple said Apple Intelligence will enable edits backed by more powerful image models, including Spatial Reframing, which lets users improve a photo’s composition after it is taken. Apple described the function as building on spatial understanding attributed to Apple Vision Pro, allowing users to touch and drag an image and preview in real time how the perspective shifts, while generating new content only in the adjusted areas so the reframed photo stays consistent with the original scene. Apple also said Extend can expand images to give subjects more room without cropping important content, and that Clean Up is getting an upgrade aimed at removing distractions with better quality infill.

Apple also said it is adopting identification for AI-altered and AI-generated images through SynthID watermarks. In its description, Photos edits made with Apple Intelligence will automatically include a hidden SynthID watermark to identify them as AI-edited. For Image Playground, Apple said a new generative model can create photorealistic images, with the generation running on Private Cloud Compute, and that generated images will also include a hidden SynthID watermark to identify them as AI-generated.

For web browsing and account protection, Apple said Safari gains new organization and monitoring tools, and its Passwords app can now help users upgrade weak or compromised credentials. Apple Intelligence-driven Safari updates described in the release include automatically organizing tabs into relevant topics, a Notify Me feature that can monitor a webpage for changes such as restocks or price drops and then notify the user, and Describe an Extension, which Apple says generates custom Safari extensions from a user’s description. In Passwords, Apple said users can automatically fix weak and compromised passwords with a tap, with the system described as securely navigating websites to sign in and upgrade accounts to strong passwords.

Apple Intelligence’s next-generation Siri, described as “Siri AI,” is positioned as a more conversational assistant with a new dedicated app and integrated tools for writing and “Visual Intelligence” across platforms. Apple said Siri can search for information across messages, emails and photos, answer questions about virtually any topic, and take action in apps. Apple said these Siri AI features are also available for developer testing starting immediately, with a beta expected for users later this year.

Beyond assistants and browsing, Apple said the update expands communication, scheduling and automation features. Messages will offer one-tap suggestions based on conversation context, including suggestions to create reminders or notes and the ability to help find photos by recognizing keywords, locations and people. Mail suggestions are described as becoming more capable with actions that can involve third-party apps, while Smart Reply can draw on a user’s personalized writing style. For scheduling, Apple said Calendar can add or modify events by describing them, and Shortcuts will add a Describe a Shortcut feature that assembles workflow steps from a natural-language request.

Apple also outlined features focused on privacy-first intelligence and accessibility, plus broader availability constraints. The architecture, according to Apple, uses on-device processing alongside Private Cloud Compute, and Apple said the next generation of Apple Foundation Models is custom-built in collaboration with Google and its Gemini models. Apple also said Private Cloud Compute is intended to ensure users’ personal data is not stored or made accessible to Apple or others when those requests are handled. For limits, Apple noted that some features including image generation have daily usage limits tied to server-based models, and it said more access is available with most iCloud+ subscription plans. Apple did not provide additional detail on how usage limits are calculated or how performance may vary by device, region or language.

Looking ahead, Apple’s launch plan sets a near-term timeline for developers to test the new features and an on-device rollout targeted for this fall through iOS 27, iPadOS 27, macOS 27, watchOS 27 and visionOS 27. What to watch next is how quickly Apple Intelligence features move from developer testing to public beta and full availability, how widely features work across supported hardware and languages, and whether independent observers view Apple’s privacy claims around Private Cloud Compute as verifiable in practice.

Image quality, watermarking and privacy claims will likely shape early user and developer reactions, while the practical impact on daily workflows will depend on how consistently Siri AI, Safari organizing and messaging suggestions behave across different input styles and use cases. Apple’s release indicates a shift toward “AI as infrastructure,” where editing, browsing and scheduling tools are expected to become more conversational and context-aware, but the company’s full user experience details are still contingent on the fall rollout.