Everything (AI) Apple announced at WWDC 2026

Apple used WWDC26 to turn Apple Intelligence into a system layer: Siri AI, Google-assisted foundation models, Private Cloud Compute, smarter apps, and new developer tools for building AI into Apple software.

Prior to now, Apple’s AI story had a very Apple-shaped problem. The company had the devices, the apps, the chips, and the right privacy pitch. But Siri still felt like the assistant you asked for to set a timer (and yelled at when it couldn't hear you to stop it) while using ChatGPT or Claude for everything that required, y'know, actual brain power.

Well, WWDC26 was Apple’s reset. The keynote’s AI 2.0 pitch was simple: Apple wants AI to become part of the operating system itself. That means an assistant that can see what you see, search your personal context, and use the apps you already rely on. It can also hand work back to the right place instead of trapping it inside a chatbot window.

The big idea: Apple is turning AI into a private routing layer for your phone, Mac, watch, headset, apps, and data. And that, my dear friends, is a very, VERY good idea...

Watch the WWDC 2026 keynote below. And for a more technical recap for how to build AI-powered apps with Apple's new AI features, check out this post.

P.S: Apple’s Newsroom press release is the cleanest landing page for the consumer-facing Apple Intelligence announcements. The keynote showed the demos; the release adds useful fine print, including SynthID watermarks, accessibility updates, availability details, and usage limits. Check it out if ya like.

The headline: Siri AI becomes Apple’s systemwide assistant

The biggest announcement was the new Siri AI. This is Apple’s rebuilt version of Siri, powered by the next generation of Apple Intelligence.

Apple described it as a much more conversational assistant that can answer richer questions, remember the thread of a back-and-forth conversation, and take action across apps. You still access it in familiar ways, like saying “Hey Siri,” but Apple also added new entry points across devices.

The practical version:

• On iPhone, you can swipe down from the Dynamic Island to search or start a typed conversation with Siri.
• On Mac, Siri is integrated into Spotlight, so the place you already use to find files and apps can now answer questions and reason over selected files.
• On iPad, Siri AI gets the same rich conversational experience.
• On Apple Watch, Siri AI gets a tailored wrist-sized experience, plus access through the new app grid.
• On Vision Pro, Siri becomes a 3D visual object you can place in space and talk to just by looking at it.
• In CarPlay and AirPods, Apple said Siri’s new capabilities extend into everyday hands-free use.

The actual shift is that Siri can now combine several kinds of context at once. In Apple’s demos, Siri found a concert date, set a reminder, identified a beach from an on-screen image, and found a friend’s address from messages. It also routed directions through that stop, searched family photos, and added only the right people to a shared album.

This is the AI people keep asking for: less “write me a haiku,” more “please make this annoying thing happen.” No six-app scavenger hunt required.

The new Siri app makes Apple’s assistant feel more like ChatGPT

Apple is also giving Siri its own dedicated app across Apple products. That matters because old Siri was mostly temporary: you asked, it answered, and the moment disappeared.

The new Siri app gives you a conversation history, lets you revisit past responses, and privately syncs those conversations through iCloud. You can start on your iPhone, continue on iPad, and finish on Mac.

This makes Siri feel more like the AI tools people already use every day. The difference is Apple’s intended advantage: Siri has system access. It can pull from your apps, your files, your messages, your calendar, your screen, your camera, and your device state.

That also makes Siri a trust test. A chatbot can be useful while knowing very little about you. A real assistant becomes useful by knowing a lot.

Apple rebuilt the architecture behind Apple Intelligence

The most important technical announcement came before the Siri demos. Apple said it built the next generation of Apple Foundation Models through a deep collaboration with Google, using technologies behind the Gemini family of models.

The distinction matters. Siri becomes an Apple interface, while the underlying Apple Foundation Models draw on Google’s Gemini technology. Apple said those models were adapted for Apple Intelligence, then deployed in two places:

• On-device, where tasks can run locally on your iPhone, iPad, or Mac.
• Private Cloud Compute, Apple’s server system for larger AI tasks that your device cannot handle alone.

Apple’s privacy pitch stayed central. Private Cloud Compute is Apple’s answer to a familiar AI tradeoff. Better models often require cloud processing, while cloud processing usually means sending personal data somewhere else. Apple says requests sent through Private Cloud Compute are used only to execute the request, are not stored, are not accessible to Apple, and can be verified by outside experts.

Apple also introduced a second, more powerful on-device model for its most capable Apple silicon systems. That version can understand and generate speech, understand text and images, improve system-wide dictation, and produce more expressive Siri voices.

In plain English: Apple is building a model stack with routing logic. It decides which AI work happens locally, which work needs Apple’s private servers, and which app or tool handles the result.

The system orchestrator is the quiet big deal

Apple also described a new system orchestrator. That sounds like platform plumbing because it is platform plumbing. It may be the piece that makes this strategy work.

The orchestrator coordinates Apple Intelligence across four ingredients:

• Personal context understanding, which helps Apple Intelligence find relevant information from your photos, notes, messages, files, and app data.
• Broad world knowledge, which lets it go to the web for current information.
• App Actions, which lets it use tools from apps to complete requests.
• On-screen awareness, which lets it understand what you are looking at and respond in the moment.

This is Apple’s version of an agentic assistant. “Agentic” means the AI can take steps toward a goal, often using tools along the way. Apple’s version is tightly tied to the operating system, which could make it more useful than a standalone chatbot for routine tasks.

The catch is that Apple needs developers to wire their apps into this system. More on that in a second.

Everything Apple Intelligence now does inside apps

Apple’s second big message was that Apple Intelligence is spreading through default apps. Some of these features are small quality-of-life upgrades. Some are much closer to agents.

Safari gets three big AI additions:

• Tab topics: Safari can analyze open pages, find similarities, and group related tabs into topics.
• Notify Me: Safari can monitor a page for changes, like tickets going on sale or a product coming back in stock.
• Describe an extension: You can describe a custom Safari extension in natural language, and Safari can create one that changes web pages for your needs.

That last one is sneaky important. It turns the browser into a programmable surface for normal people. Someone who wants a recipe-saving button on food blogs should be able to ask for it in plain English.

Passwords gets an even more agent-like feature. If Passwords finds weak or compromised passwords, it can use Apple Intelligence and Safari to update eligible accounts with one tap. Apple said it can navigate websites, sign in, and change the password on your behalf.

Messages, Mail, Calendar, and Phone are getting contextual suggestions:

• Messages can suggest one-tap actions like creating reminders or notes from a conversation.
• Messages can also help find the right photos when someone asks for them.
• Mail can surface more capable action suggestions based on the message context, including actions from third-party apps.
• Calendar can create or edit events from natural language.
• Phone gets Call Context, which surfaces relevant information from your apps when you call a business.

The Phone example is very Apple. If you call an airline, your confirmation code can appear automatically from Mail. Apple emphasized that the feature looks at who you are calling, rather than what you are saying, and runs on device.

Visual Intelligence moves into the camera, screen, and Vision Pro

Visual Intelligence is Apple’s name for AI that can understand images, screens, and objects around you.

On iPhone, it now lives in the Camera app through a new Siri mode. You can point the camera at something, tap the shutter, and ask Siri about it. Apple showed examples like nutrition insights from a plate of food and splitting a bill by selecting what you ordered, then using Apple Cash.

On Mac, Visual Intelligence gets a keyboard shortcut. You can select part of your display and ask Siri about it. If you are looking at a schedule, for example, Siri can suggest adding multiple events to Calendar.

On iPad, Visual Intelligence is built into screenshots. On Vision Pro, it extends into spatial computing: you can look at something in your physical or digital space and ask Siri about it.

This is where Apple’s platform advantage becomes obvious. The company does not need to build a separate AI camera app, AI screen reader, AI assistant, and AI headset feature. It can make Siri the same interface across all of them.

Writing Tools grow from “rewrite this” to “help me work”

Apple is also expanding Writing Tools into the Siri experience.

The big upgrades:

• Siri can draft from scratch anywhere you type.
• In Mail and Messages, Siri can reflect how you usually communicate with a specific person.
• You can select writing and ask Siri for feedback.
• Automatic proofreading is available system-wide, including in most third-party apps.

The style-matching detail is the one to watch. Apple is saying Siri can adapt to the relationship, so an email to your manager can come out as quick bullets while a message to a friend can sound more casual.

That is useful, and also delicate. If Apple gets it right, it saves time. If it gets it wrong, everyone gets work emails that sound like a hotel concierge got trapped in Outlook.

Image Playground gets real image generation

Apple’s first Image Playground era was more cute than essential. WWDC26 changes the pitch.

The new Image Playground uses a more powerful generative model running on Private Cloud Compute. Apple says it can create high-quality images in almost any style, including photorealistic images. Apple’s Newsroom release says generated images automatically include a hidden SynthID watermark that identifies them as AI-generated.

The feature can start from a photo or text description, include people from your Photos library, transform photos into different styles, and refine images through natural language or touch. You can also choose dimensions, like landscape for a website or portrait for a flyer.

Apple is integrating Image Playground into more system surfaces too, including Messages backgrounds, contact posters, and Lock Screen wallpapers. Developers can tap into the same capabilities through the Image Playground API.

Photos gets Clean Up, Extend, and Spatial Reframing

Photos received one of the more interesting AI feature sets.

Apple announced three editing tools:

• Clean Up gets better object removal and more realistic infill.
• Extend expands an image beyond its original edges, useful for fixing aspect ratios or giving a subject more space.
• Spatial Reframing lets you shift the apparent camera position after the photo was taken.

Spatial Reframing is the most Apple-y demo of the group. It combines on-device spatial models with a generative model on Private Cloud Compute. You drag the photo as if you had moved the camera in the original scene. The system previews the perspective change locally, then generates only the missing edge content to finish the image.

Apple said these tools work on almost any photo in your library, including older photos and photos taken with other cameras. Apple’s Newsroom release also says photos adjusted with Apple Intelligence automatically include a hidden SynthID watermark that identifies them as AI-edited.

Home, cameras, and Shortcuts get more useful

Apple Intelligence is also coming into Home and Shortcuts.

In Home, Apple Intelligence can group related accessory notifications into one activity. Instead of getting a pile of separate alerts, you get one notification that updates as the event unfolds.

Home cameras get more meaningful AI too. The Home app can analyze recorded clips from compatible cameras, summarize what happened, pull together related footage from multiple cameras, and let you search clips by what was captured. Apple also said supported cameras can show recorded clips in 4K.

Shortcuts gets a major accessibility upgrade. Instead of manually connecting steps, you can describe what you want in natural language. Apple’s example was simple: “When I’m leaving work, message Pedro, ‘I’m on my way’ with my ETA.” Shortcuts turns that into an automation. It detects when you leave work, calculates your ETA through Maps, and sends the message.

That is exactly where Shortcuts needed help. Power users loved it. Normal people saw a wall of blocks and quietly backed away like they had opened the electrical panel at an Airbnb.

Accessibility and smaller system features also got Apple Intelligence

Apple’s Newsroom release also calls out AI-powered accessibility updates. VoiceOver can give richer image descriptions, Live Recognition can answer questions about surroundings, Magnifier brings similar assistive exploration to a high-contrast interface, Voice Control can understand described onscreen buttons, and Accessibility Reader can summarize and translate more complex source material.

The release also lists smaller system upgrades: automatic proofreading across the system, intelligent file and folder name suggestions based on contents, better Genmoji quality with natural-language edits, and Workout Buddy support in Spanish and on Apple Watch when the iPhone is not nearby.

Developers get the real platform story

For developers, WWDC26 was less about one consumer feature and more about Apple building an AI toolchain.

The Foundation Models framework is the centerpiece. It gives developers a native Swift API for Apple’s on-device model. It now supports more model options. Apple says developers can work with Apple Foundation Models, cloud models like Claude and Gemini, or any provider that conforms to Apple’s Language Model protocol.

That is a big positioning move. Apple is saying developers can build AI features inside native apps without treating every model as a separate integration project.

The framework also adds:

• Multimodal prompts, so apps can pass images alongside text.
• Vision framework tools, like OCR and barcode reading, available for models to call.
• Dynamic Profiles, which let apps swap models, tools, and instructions during a continuous session.
• Evaluations, so developers can test whether AI features behave correctly across changing conditions.

Apple also said smaller apps can access the latest Apple Foundation Model on Private Cloud Compute. The Apple Intelligence guide says this applies to apps in the App Store Small Business Program with fewer than 2 million total first-time downloads. The guide says there is no cloud API cost.

App Intents are how third-party apps get into Siri AI

App Intents may be the most important developer framework in the whole announcement.

The App Intents framework lets developers connect app content and actions to Apple Intelligence. In normal terms: it tells Siri what your app contains and what your app can do.

Two pieces matter:

• Entity schemas add app content to the Spotlight semantic index, so Siri can find relevant information inside third-party apps.
• Intent schemas let Siri take action in those apps through natural language.

Apple says developers do not need to define exact trigger phrases. That is important because old voice assistant integrations often broke when users failed to say the magic words.

The new View Annotations API adds on-screen awareness, so developers can map interface elements to entities. If you are looking at something in an app, Siri can understand what “this” refers to.

If this works, Siri AI becomes less like a chatbot and more like a universal command layer for apps. If developers do not adopt it well, Siri stays smartest inside Apple’s own apps.

Core AI, MLX, and Xcode show Apple wants the full AI dev stack

Apple also announced Core AI, a framework for running developers’ own models on device. It is built into the OS, optimized for Apple silicon, and designed for private, local execution with zero server dependencies and zero token costs.

That matters for companies that want AI features but do not want every prompt leaving the device.

The AI & Machine Learning guide also highlights MLX, Apple’s open-source framework for Apple silicon. MLX is getting support for Metal 4 and GPU Neural Accelerators, plus distributed training across multiple Macs using RDMA over Thunderbolt.

Xcode gets its own agentic coding upgrades. Apple’s Platforms State of the Union says agents in Xcode can run tests, try things in Playground, run apps in the Simulator, fix issues, and localize apps. Developers can extend agents with custom skills, connect MCP tools, and use the Agent Client Protocol.

Apple also said Xcode’s coding assistant can connect to tools like Figma and GitHub and use model and agent options, including Gemini.

The theme is clear: Apple wants AI development to feel native on Apple platforms, from model runtime to app actions to coding agents.

Availability, limits, and the fine print

Apple’s rollout has several caveats.

Siri AI will be available in English first, with more languages coming later. Developers can start trying the new Siri now, and customers will get Siri AI in beta later this year.

The broader software releases are developer betas now, public betas next month, and public releases this fall.

Apple’s Newsroom availability section has the full supported-language and compatible-device list for iOS 27, iPadOS 27, macOS 27, watchOS 27, and visionOS 27. It also notes that Suggestions in Mail and Messages, Call Context, and intelligent file and folder naming will be available in English.

Apple said the update supports the same product models that support Apple Intelligence today. The most powerful on-device model and its features, like expressive Siri voices and more advanced dictation, are limited to Apple’s most capable iPhone, iPad, and Mac systems.

Apple also said Siri AI will not be available initially in the EU on iOS and iPadOS while it works on a path that preserves privacy and security. In China, Siri AI and the other new Apple Intelligence features will be unavailable while Apple works through regulatory requirements.

There are also usage limits. Apple said some features, including image generation, have daily usage limits because they rely on powerful server models. Increased access is available with most iCloud+ subscription plans, which also include Apple Intelligence support for compatible home cameras.

Reading between the lines

The easy reaction is that Apple is catching up. A more conversational Siri, image generation, browser tab grouping, writing help, AI photo edits, and natural language automations all have obvious cousins elsewhere.

That critique is fair. The Verge called some of the Siri AI feature set familiar compared with Gemini on Android and third-party chatbots. Apple also carries baggage from the earlier Apple Intelligence rollout, where promised Siri improvements took longer than expected to arrive.

Apple’s counterargument is distribution and trust. ChatGPT, Claude, and Gemini are destinations. Apple wants AI to become infrastructure: inside the keyboard, camera, browser, calendar, phone app, Photos library, Spotlight, Shortcuts, and third-party apps.

That is a different bet. Apple is not trying to make users live in one AI app. It is trying to make every app a little more actionable through the same assistant layer.

The unresolved issue is execution. Siri AI only becomes compelling if it works reliably across the messy middle of real life. That means half-remembered contacts, badly named files, third-party apps, blurry screenshots, weird websites, incomplete calendars, and requests that require judgment.

Apple has the hardware and software stack to attempt that. It also has a painful place to fail: the assistant that hundreds of millions of people already learned to underuse.

What this means for you

For regular users, WWDC26 means Apple Intelligence is moving into everyday device behavior. The most useful changes will probably feel small at first. Think better dictation, automatic proofreading, smarter photo search, Safari monitoring a page, Passwords fixing logins, Calendar understanding plain English, and Siri finding something buried on your device.

For developers, the work starts sooner. App Intents, Foundation Models, Core AI, Evaluations, and Xcode agents are the pieces to watch. Apple is giving developers a path to make apps available to Siri AI, run models locally, test AI behavior, and connect their own tools into the platform.

For businesses, the bigger implication is privacy-shaped distribution. Apple is offering a way to bring AI closer to sensitive data without automatically sending that data into a generic cloud chatbot.

That is the story of Apple’s WWDC26 AI announcements: less spectacle, more plumbing. Siri got the headline. The real bet is that AI becomes useful when it lives where the work already happens.

Keynote timecode index

• Apple turns to intelligence: Craig opens the Apple Intelligence section.
• New Apple Intelligence architecture: Apple frames the architecture behind the next generation of Apple Intelligence.
• Google Gemini collaboration: Apple says it worked with Google on the next generation of Apple Foundation Models.
• Private Cloud Compute deployment and privacy promise: Apple explains where the models run and how requests are handled.
• System orchestrator: Apple describes the coordination layer across personal context, web knowledge, app actions, and screen awareness.
• Siri AI introduction: Apple names its rebuilt assistant.
• Siri world-knowledge demo, on-screen awareness demo, and personal-context demo: Siri combines web knowledge, screen context, and messages.
• Siri Photos demo: Siri finds family photos and adds selected people to a shared album.
• Expressive Siri voice and improved dictation: Apple shows the advanced on-device model features.
• Conversational Siri on iOS: Siri supports richer back-and-forth conversations.
• Watch party planning demo: Siri plans a menu, pulls personal context, and drafts a group message.
• Siri on Mac and Spotlight: Siri moves into Spotlight and file context menus.
• Siri compares files and drafts an email: Siri reasons over documents and writes from scratch.
• Dedicated Siri app: Apple introduces conversation history synced through iCloud.
• Siri on Apple Watch and Siri on visionOS: Siri expands across more Apple platforms.
• Visual Intelligence begins: Siri gets visual understanding across camera, screen, and Vision Pro.
• Camera Siri mode, Mac Visual Intelligence, iPad screenshot integration, and visionOS visual Q&A: Apple shows the platform-specific visual entry points.
• Writing Tools with Siri: Siri can draft, adapt tone, give feedback, and proofread system-wide.
• Apple Intelligence inside apps: Apple moves from Siri to app integrations.
• Safari tab topics, Notify Me, and Describe an extension: Safari gets AI organization, monitoring, and custom extensions.
• Passwords auto-updates logins: Passwords uses Apple Intelligence and Safari to update eligible passwords.
• Messages suggestions, Mail suggestions, Calendar natural language events, and Phone Call Context: Apple shows communication and scheduling features.
• Home notifications, Home camera summaries, and Shortcuts from natural language: Apple adds intelligence to smart home and automation workflows.
• Image Playground upgrade: Image Playground gets higher-quality generation and photorealistic styles.
• Photos AI editing begins, Clean Up, Extend, and Spatial Reframing: Photos gets new AI editing tools.
• Usage limits and iCloud+ access: Apple explains server-model limits and expanded access.
• Siri AI beta timing, EU availability caveat, and China availability caveat: Apple explains rollout limits.
• Developer tools begin: Apple turns to App Intents, Foundation Models, Core AI, and Xcode.
• App Intents examples, Foundation Models framework, Core AI, Xcode coding assistant, and Device Hub: Apple outlines the developer AI stack.

Key Moments from the Keynote, with Timecodes

• (0:00:27) Tim opens WWDC by framing it as Apple’s annual moment to share platform advancements with developers and users.
• (0:00:53) Tim says developer energy around Apple platforms “has never been stronger,” with “well over 1,000 submissions to the App Store every hour.”
• (0:01:06) Apple says it is investing in the next generation through 20 Apple Developer Academies that have helped tens of thousands of students.
• (0:01:26) Tim restates Apple’s product philosophy: technology should be “personal, powerful, and easy to use,” with hardware and software tightly integrated.
• (0:02:16) Tim teases “latest advancements in Apple Intelligence and Siri” as a central part of the keynote.
• (0:02:43) Craig says this year’s releases focus on three areas: responsiveness and usability, trust and safety for kids, and a major Apple Intelligence / Siri upgrade.
• (0:04:48) Apple names the next macOS release “macOS Golden Gate.”
• (0:05:23) Craig says Apple is focusing on improving features users already rely on, arguing the best operating systems come from “sweating the details.”
• (0:06:45) Shubham says Apple refined Liquid Glass after listening to user and developer feedback.
• (0:07:13) Apple updated the foundations of how Liquid Glass is built.
• (0:07:20) Liquid Glass now diffuses complex content behind it more effectively to improve readability, depth, and separation.
• (0:07:30) Apple is adding a Liquid Glass slider so users can choose anything from “ultra clear” to “fully tinted,” and developer apps that already adopted Liquid Glass get the customization automatically.
• (0:08:09) macOS gets a more uniform toolbar across app tops to make controls, labels, and headings easier to read.
• (0:08:24) macOS sidebars now expand to the edge of windows to reduce distractions and make apps feel more expansive.
• (0:08:41) Sidebar icons regain color so users can distinguish apps and identify the foreground window more easily.
• (0:08:48) Every macOS window now uses the same tighter corner radius for consistency, including apps that have not been updated.
• (0:09:01) Apple is adding additional Liquid Glass layers into app icon artwork so icons appear sharper and more defined.
• (0:09:56) Stacey says Apple optimized fundamentals like memory usage, CPU utilization, networking, and display rendering to improve responsiveness.
• (0:10:13) System animations are smoother, including swiping between Home Screen pages, entering Mission Control, and moving between spaces on Mac.
• (0:10:23) iPhone and iPad apps launch up to 30% faster because Apple preloads key data apps need when opened.
• (0:10:50) New photos appear in the Photos library up to 70% faster.
• (0:11:00) AirDrop transfers are up to 80% faster.
• (0:11:07) Browsing and transferring files from iPad to an external drive is up to five times faster, bringing it in line with Finder on Mac.
• (0:11:17) Apple is bringing its advanced CPU scheduler improvements to older iPhones, going back to iPhone 11.
• (0:12:03) iOS 27 supports iPhone 11 and all the same iPhone models as iOS 26, which Apple says makes it available to more users than any iOS release ever.
• (0:12:33) Apple says iPhone network transitions between cellular and Wi-Fi are smarter, reducing moments where users manually toggle Wi-Fi off.
• (0:13:06) Messages gets send indicators for slow outgoing messages, and large photo or video sends should no longer slow the whole conversation.
• (0:13:35) Apple rebuilt the search foundation powering Spotlight, Photos, and Mail.
• (0:13:50) The new search index is designed to be more stable, efficient, and comprehensive across old and new content.
• (0:14:22) Mail gets a new ranking system so more relevant results appear in Top Hits, even for older emails.
• (0:14:56) iCloud Shared Albums can now include photos and videos contributed by friends on Android or Windows, with full-resolution sharing.
• (0:15:09) Health adds Cycle Tracking support for perimenopause and menopause, including notifications, symptom logging, and educational information.
• (0:15:34) AirPods get custom EQ for more personalized sound.
• (0:15:49) Apple Vision Pro can turn panoramas into spatial scenes and use those panoramas as immersive environments.
• (0:16:07) Maps Flyover gets sharper city detail by combining aerial imagery with vision intelligence models.
• (0:18:04) Sumbul says Apple’s child safety work rests on two principles: every child is unique, and parents are best positioned to decide what works for their family.
• (0:18:33) Apple says expert research emphasizes balancing technology with in-person social interaction, academics, physical activity, and sleep.
• (0:18:45) Apple says kids under 18 benefit from age-based protections with adult supervision, and under-13 device access should be limited and expanded as a child becomes ready.
• (0:19:06) Apple is working with the American Academy of Pediatrics to adapt the Family Media Plan into a parent guide for Apple child safety features.
• (0:19:49) Raja says the first step is creating a Child Account, which immediately enables age-tailored safeguards across the system.
• (0:20:18) Apple says this year’s child safety focus is what kids can see, who they can talk to, when they have access, and how parents guide their digital journey.
• (0:21:04) Parents can start kids with only content they are comfortable with, then gradually add more over time.
• (0:21:17) Setup Assistant lets parents choose a few essential apps, a recommended set, or specific apps for a child.
• (0:22:08) Apple introduces Ask to Browse, expanding parental approval from app downloads to new websites in Safari.
• (0:22:32) Ask to Browse and Ask to Buy are on by default for children under 13, and parents can enable them for teens.
• (0:23:01) Communication Safety warns children and blurs images or videos that may contain nudity, including on live FaceTime calls.
• (0:23:14) Communication Safety will also intervene before kids see gore or violent content in shared images or videos.
• (0:23:47) Screen Time adds Time Allowances, with recommendations for Entertainment, Games, and Social Media.
• (0:24:12) Apple says Time Allowance recommendations are age-based and developed with clinical and child development experts, including the American Academy of Pediatrics.
• (0:24:32) Apple says experts recommend children under 13 avoid social media, while parents should carefully consider when teens are ready.
• (0:24:56) Parents can set schedules so kids have different app access at different times, such as school days versus weekends.
• (0:25:40) Screen Time has been redesigned so parents can see usage at a glance and adjust access quickly.
• (0:26:12) Apple says developers share responsibility for age-appropriate experiences inside apps, since app experiences can include AI, immersive gaming, and social features.
• (0:26:54) The Declared Age Range API lets developers use a child’s age range in a privacy-preserving way to tailor app experiences.
• (0:27:53) Craig frames AI as powerful technology that can shape society and unlock meaningful benefits “with proper care.”
• (0:28:04) Craig contrasts Apple’s approach with companies “seemingly pursuing AI, for the sake of AI,” saying Apple wants AI centered on people.
• (0:28:31) Apple says helpful AI must be grounded in personal context, the apps users rely on, and privacy at every step.
• (0:29:25) Apple Foundation models are described as the heart of Apple Intelligence.
• (0:29:30) Apple says it collaborated deeply with Google, using technologies behind Gemini models to create the next generation of Apple Foundation models.
• (0:29:45) The new Apple Foundation models are adapted to run both on-device and on servers through Private Cloud Compute.
• (0:29:51) Apple says the models unlock stronger understanding, reasoning, and multimodal abilities, including image understanding and generation.
• (0:30:13) Apple built a second, more powerful on-device model for its most capable Apple silicon systems, supporting speech understanding, speech generation, better dictation, and expressive voices.
• (0:30:39) Apple Intelligence uses a new system orchestrator to securely coordinate across models and system-wide capabilities.
• (0:30:53) Personal context understanding uses Spotlight and a semantic index to surface relevant user content.
• (0:31:17) Apple Intelligence can use broad world knowledge by going to the web for up-to-date information, then generating answers through Private Cloud Compute.
• (0:31:31) App Actions let Apple Intelligence use tools from apps to complete requests, such as drafting emails or editing and sharing photos.
• (0:31:45) On-screen awareness lets Apple Intelligence tailor assistance based on what app the user is in and what they are doing.
• (0:32:39) Apple says Apple Intelligence uses on-device processing and Private Cloud Compute, where user data is only used to execute the request and is not stored or accessible to Apple.
• (0:33:26) Mike introduces Siri AI, a rebuilt Siri with “powerful AI at the core.”
• (0:33:48) Siri AI combines personal context understanding, app actions, on-screen awareness, image understanding, and broad world knowledge.
• (0:34:17) Siri AI is more conversational, and users can refer back to conversations through a dedicated Siri app.
• (0:34:31) Siri gains Visual Intelligence across platforms plus integrated tools to write and edit almost anywhere users type.
• (0:35:09) Demo: Siri answers a current-world-knowledge question about a Suki Waterhouse show, then helps with tickets, a reminder, and music playback.
• (0:36:06) Demo: Siri uses on-screen awareness to identify a Santa Cruz coast location in a photo.
• (0:36:17) Demo: Siri finds a friend’s new address from personal context, even though the user did not save it manually.
• (0:36:35) Demo: Siri combines world knowledge and personal context to create directions to a landmark with a stop at a friend’s place.
• (0:37:01) Demo: Siri finds photos from a family trip and adds only photos with named people to a shared family album.
• (0:37:39) Apple says its most advanced on-device model enables a more expressive Siri voice experience.
• (0:38:12) Users can customize Siri’s voice expressivity and pace.
• (0:38:38) System-wide dictation gets higher accuracy for spelling, punctuation, and capitalization.
• (0:39:15) Justin says Siri AI can now handle richer conversations, including in-depth plans, brainstorms, and document feedback.
• (0:39:40) On iOS, users can swipe down from the Dynamic Island to search or start typing a Siri conversation.
• (0:39:56) Demo: Siri answers a World Cup schedule question, then expands into a richer conversational planning interface.
• (0:41:18) Demo: Siri combines World Cup context, iconic dishes from Brazil and Morocco, and a family member’s coconut cookie message into a watch-party menu.
• (0:41:51) Demo: Siri drafts and sends a message to a group chat with the watch-party menu included.
• (0:42:32) On macOS, Siri is integrated into Spotlight so users can ask questions from anywhere on the Mac.
• (0:42:47) macOS context menus add a Siri field so users can ask about selected images, files, or text.
• (0:44:04) Demo: Siri compares multiple shed quote files and produces a table explaining how they stack up.
• (0:44:44) Demo: Siri handles a misspelling, searches Messages and emails for a son’s electrical issue, and uses that context to recommend a shed option.
• (0:45:17) Demo: Siri drafts an email to a contractor, extracting the contractor’s name and email while explaining why the user chose them.
• (0:45:50) Siri AI’s conversational experience is coming to iPadOS, with a dedicated Siri app across Apple products.
• (0:46:12) The Siri app shows conversation history that syncs privately through iCloud across iPhone, iPad, and Mac.
• (0:46:26) watchOS gets Siri AI so users can ask questions and act from their wrist.
• (0:46:44) visionOS gets a 3D visualization of Siri that users can place in space and speak to by looking at it.
• (0:47:24) On iPhone, Visual Intelligence is integrated into the Camera app through a new Siri mode.
• (0:47:49) Siri’s deep image understanding is powered by Apple Foundation models running on Private Cloud Compute.
• (0:47:57) Camera-based Siri mode suggests relevant actions based on what is in front of the user.
• (0:48:02) Example: users can point iPhone at food to get nutritional insights.
• (0:48:07) Example: users can point iPhone at a bill, select what they ordered, and split the tab with Apple Cash.
• (0:48:15) macOS gets Visual Intelligence through a keyboard shortcut, letting users select part of the display and ask Siri about it.
• (0:48:40) Visual Intelligence can suggest actions from on-screen content, such as adding multiple events from a schedule to Calendar.
• (0:48:50) iPad Visual Intelligence is integrated into the screenshot experience.
• (0:49:00) On visionOS, users can ask Siri about things by looking at them, including physical objects in the room.
• (0:49:41) Siri Writing Tools can generate drafts from scratch anywhere users type.
• (0:50:02) In Mail and Messages, Siri can reflect how the user usually communicates with a specific colleague or friend.
• (0:50:16) Users can select writing and ask Siri how it sounds, getting feedback and suggestions.
• (0:50:26) Apple Intelligence automatically proofreads as users type across the system, including most third-party apps.
• (0:51:18) Siri AI launches in English first and will “quickly expand” to more languages.
• (0:52:41) Safari uses Apple Intelligence to organize tabs into topics and keep adding related tabs to the right topic as users browse.
• (0:53:16) Safari Notify Me can monitor a page for updates, such as camp signups or product restocks.
• (0:53:35) Users can describe in natural language what Safari should watch for, then close the tab and wait for a notification.
• (0:53:47) Apple says Safari’s AI features do not share sensitive browsing data with anyone, “not even Apple.”
• (0:54:08) Safari adds “Describe an extension,” which lets users create custom extensions by describing what they want in natural language.
• (0:54:30) Passwords uses Apple Intelligence to help users upgrade weak or compromised passwords.
• (0:54:44) Passwords can agentically navigate websites through Safari to sign in and update eligible accounts to strong passwords.
• (0:55:20) Messages uses Apple Intelligence to understand conversation context and suggest one-tap actions like reminders or notes.
• (0:55:33) Messages can help find requested photos by recognizing keywords, locations, and people in the user’s library.
• (0:55:48) Mail offers more capable contextual suggestions, including actions through third-party apps.
• (0:55:59) Calendar can create events from natural language, identifying contacts, locations, and titles while the user types.
• (0:56:16) Calendar can edit events from natural language, such as changing a weekly team lunch to every other week.
• (0:56:25) Phone adds Call Context, surfacing relevant information from apps when calling a business.
• (0:56:48) Apple says Call Context looks at who the user is calling, not what they are saying, and runs entirely on device.
• (0:57:16) Home uses Apple Intelligence to group related accessory notifications into a single updating activity.
• (0:57:46) Home can analyze recorded clips from compatible cameras and generate descriptions of what happened.
• (0:57:55) Home can connect relevant footage from multiple cameras so users see the whole activity across clips.
• (0:58:11) Home camera search can find clips by what was captured and elevate important clips before the user types.
• (0:58:37) Home adds 4K recorded clips on supported cameras.
• (0:58:50) Shortcuts becomes more approachable by letting users describe a shortcut in natural language.
• (0:59:20) Shortcuts uses Apple Intelligence to reason over the user’s description and assemble the required steps.
• (0:59:34) Example: a user can describe an automation that messages their partner with an ETA every time they leave work.
• (0:59:53) Users can revise Shortcuts by describing changes in natural language, such as adding podcast playback.
• (1:00:17) Image Playground is being rebuilt around more powerful image models.
• (1:00:30) Image Playground can generate high-quality images in many styles, including photorealistic images, through a Private Cloud Compute model.
• (1:00:50) Image Playground can make images using multiple people from the user’s Photos library.
• (1:00:57) Users can transform photos into many styles by describing what they want in natural language.
• (1:01:04) Apple says photos used for Image Playground are never stored or shared, even with Apple.
• (1:01:13) Users can start with a person from Photos and a description to create a custom image, such as a birthday invitation.
• (1:01:29) Image Playground lets users refine images by describing changes in natural language.
• (1:01:38) Users can circle objects to select them, then move, resize, or modify them with natural language.
• (1:01:52) Image Playground supports dimensions for different use cases, such as landscape images for websites and portrait images for flyers.
• (1:02:05) Image Playground is integrated across the system, including Messages backgrounds.
• (1:02:12) Apple is bringing Image Playground into contact posters and Lock Screen wallpapers.
• (1:02:19) Image Playground can suggest images inspired by the user’s photos, favorite locations, and activities.
• (1:02:27) Developers get access to the new Image Playground features through the Image Playground API.
• (1:02:44) Apple says its goal for AI in Photos is to enhance images while respecting the original moment.
• (1:03:20) Photos Clean Up gets better-quality removal and more realistic infill, including in complex scenes.
• (1:03:34) Photos adds Extend, which expands images or straightens crooked horizons without cropping out important content.
• (1:03:57) Apple introduces Spatial Reframing, a Photos tool that builds on Clean Up, Extend, and spatial models from Apple Vision Pro.
• (1:04:23) Spatial Reframing lets users fix photo composition after capture, such as shifting perspective or improving framing.
• (1:05:09) Users can directly manipulate a photo, and the perspective shifts as if the camera had moved in the original scene.
• (1:05:18) Spatial Reframing uses on-device spatial models so users can preview changes in real time.
• (1:05:25) As users reframe, blurred edges mark areas that generative models will later fill in.
• (1:05:44) Spatial Reframing combines on-device spatial models with an image generation model running on Private Cloud Compute.
• (1:05:55) Apple says Spatial Reframing only generates new content in the gaps created by the perspective shift, keeping the photo consistent with the original scene.
• (1:06:33) The new Photos AI features work on almost any photo in a user’s library, including older photos and photos taken with other cameras.
• (1:07:01) Apple says the new Apple Intelligence app features will be free with the latest software releases.
• (1:07:05) Some features, including image generation, have daily usage limits because they rely on server models.
• (1:07:12) Increased access is included with most iCloud+ plans, which also include Apple Intelligence support for compatible home cameras.
• (1:07:54) The Apple Intelligence update supports the same product models that already support Apple Intelligence today.
• (1:08:00) The most powerful on-device model and features like expressive voices and more advanced dictation come to Apple’s most capable iPhone, iPad, and Mac systems.
• (1:08:11) Developers can start trying the new Siri immediately, while customers get Siri AI in beta later this year.
• (1:08:23) Siri AI will not initially be available in the EU on iOS and iPadOS while Apple works on a privacy- and security-preserving path forward.
• (1:08:34) Siri AI and other new Apple Intelligence features will not be available in China while Apple works through regulatory requirements.
• (1:08:51) Developers can bring Apple Intelligence into apps using App Intents so users can get information and take action through Siri.
• (1:09:25) The Foundation Models framework now supports image input in addition to text.
• (1:09:41) Developers can extend model capabilities with custom skills and use server-hosted models through the same Swift API.
• (1:10:19) Apple introduces Core AI, a framework that lets developers run other models locally in apps using Apple silicon.
• (1:10:50) Xcode’s coding assistant can localize an entire app, interact with simulated devices, and be extended with custom skills.
• (1:11:12) Xcode developers can choose the model and agent they want, including Gemini, and connect to tools like Figma and GitHub.
• (1:11:21) Apple introduces Device Hub, a unified interface for real and simulated devices with multitouch simulation, appearance switching, and dynamic resizing.
• (1:12:51) Tim says the new OS releases are available as developer betas today, with a public beta next month and user availability this fall.