Federico Viticci (Mastodon):

The AI frontier from the two big labs (and – we have to be honest with the numbers here – probably SpaceXAI and Meta now, too) is in a completely different league at the moment. I spent my first couple of weeks with Siri AI trying to convince myself that I could replace Codex with Siri AI, and that was the wrong approach. There is a potential future in which Siri AI can become a more proactive and capable agent, but that’s not what Apple is shipping this year.

[…]

Siri AI is a dramatically better Siri for typical Siri stuff. The new Siri understands your commands and nuanced requests better because of the LLMs behind it; it can often perform multiple requests in one go (such as setting multiple timers with one command); the new expressive (and customizable!) voices sound great, and they’re terrific at reading and summarizing notifications. Apple’s first advantage compared to other chatbots is right here: only Siri can integrate with the iPhone and iPad to this extent, so even if the Foundation models aren’t as smart as the competition, Siri can do a lot more for most people on their devices because neither OpenAI nor Anthropic have OSes they can control.

[…]

The biggest issue I’ve encountered is the inconsistency of it all. Very often, the new Siri AI still replies with a generic “Uh oh, something went wrong“ error that doesn’t explain what the error was or why it happened. Sometimes, I ask a question via voice, and it errors out, but then a few seconds later, I type the same question in the app, and it’s fine.

[…]

My problem is that, right now with Apple’s own apps, both the index and Siri’s ability to perform actions are very inconsistent. The assistant hallucinates abilities it doesn’t have, and it frequently just doesn’t work.

Dan Moren:

The good news is that Siri AI is capable: impressively so, at times. In the lowest of low bars, it certainly beats the pants off the previous incarnation of Siri, but its access to your own personal data and all of the app on your phone also unlocks a host of new possibilities.

But it’s also not without the pitfalls that underlie all LLM-powered chatbots: errors in understanding, hallucinations, and so on.

[…]

The app itself is pretty barebones, letting you view those conversations in List or Grid layouts, select multiple conversations to delete them, and pin or rename any single conversation. In Settings : Siri, you can choose to either open to the overview or to a new conversation and decide whether old conversations are kept for 30 days, 1 year, or forever. It’s a little strange that one-off queries (“What’s the average height of the Norwegian World Cup team?”) are stored alongside more involved conversations (“Can you help me build a specific RSS feed for this site?”); I also ended up with a few orphan blank “New conversations” which I deleted, but it’d be nice if you could filter to find those. Conversations are fully searchable and sync between devices via iCloud.

John Voorhees:

Despite the current limitations, if you use chatbots as a replacement for Google Search, you’ll probably have a good experience with the new Siri app. That’s because even in these early betas, if you ask Siri something like, “Give me a day trip itinerary for a trip to Chicago this Saturday,” you’re likely to get good but pretty middle-of-the-road recommendations. From there, though, you can drill down with follow-up queries to get something more in line with your tastes. The results include source links you can follow, too. Given the ad-strewn mess that web search has become, using Siri for this sort of thing is a meaningfully better experience.

The one caveat on Siri’s world knowledge is that it’s roughly 2-3 weeks stale in some places.

[…]

A good example is Music. You can prompt Siri to play a song, but most of the rich metadata in the app isn’t exposed to Siri, which is a shame because it is exposed elsewhere, for example to AppleScript.

Previously:

• macOS 27 Public Beta
• iOS 27 Public Beta
• No Siri AI in EU
• Siri AI Announced

Update (2026-08-04): Benjamin Mayo:

Thanks to the top-to-bottom revamp that is Siri AI, Siri is reliably intelligent. Finally. We are still in the midst of beta season, and Apple is yet to confirm whether Siri AI will formally launch with iOS 27 in September or sometime later in the year, but the feature is up to scratch. There are bugs and glitches and edge-case quirks, but on the whole, it wipes the floor with old Siri and is competent enough to supplant my everyday usage of other chatbots too.

[…]

Siri AI can get a snapshot of what’s on screen, but it cannot control the UI for you autonomously. As such, if you ask it to do something outside of its library of available app actions, it will say it can’t do that. Often, it can quote the relevant steps for you to do it yourself, but obviously that falls short of the ideal experience. This feels a bit lacklustre to people already embracing AI agents en masse. Claude and Codex on Mac do not hit these walls; they are all too keen to move your mouse for you and run terminal commands until the job is done.

Previously:

• ChatGPT Work and ChatGPT Classic
• Claude Desktop App
• Perplexity Personal Computer
• Codex for Almost Everything

Sarah Perez (PDF, Hacker News, MacRumors):

The iPhone maker alleges that this misconduct, which it says reveals a pattern of theft from OpenAI employees who previously worked at Apple, was directed by OpenAI’s senior leadership, including Chief Hardware Officer Tang Tan.

The lawsuit, which was filed in the U.S. District Court for the Northern District of California, accuses Tan of using Apple’s confidential project code names during OpenAI’s recruiting process, asking job candidates to bring in Apple hardware components to their interviews, coaching departing Apple employees on how to evade the company’s security procedures, and asking for details about the company’s unannounced products.

Apple’s allegations are detailed and shocking, and it sounds like they have a lot of evidence.

Chance Miller:

The lawsuit names Chang Liu and Tang Tan as two of the defendants. Tang Tan served as VP of product design at Apple, leading iPhone and Apple Watch product design. He departed the company in February 2024 to work with Jony Ive. Chang Liu, meanwhile, worked at Apple for eight years and was a senior system electrical engineer before departing to join OpenAI in January 2026.

Apple’s lawsuit also names OpenAI and io Products as defendants.

OpenAI’s hardware efforts are being led by Jony Ive, Apple’s former chief design officer.

[…]

Ive, Hankey, and Cannon are not personally mentioned anywhere in Apple’s initial filing today.

Emma Roth:

Here are the most surprising claims in Apple’s 41-page filing.

Marcus Mendes (MacRumors):

OpenAI’s Director of Strategic Communications, Drew Pusateri, took to X to comment on the company’s behalf.

In his post, he claims OpenAI has “no interest in other companies’ trade secrets,” adding that OpenAI remains “focused on building innovative technology that empowers people everywhere.”

This seems like a weak response, given that Apple says it reported its findings to OpenAI in February and never heard back. Presumably, they’ve already investigated but see the situation as serious enough that they need to say as little as possible.

It’s not clear to me the extent to which Apple believes the behavior was limited to the named defendants. It’s hard to believe that the rest of OpenAI’s leadership—including Altman and Ive—were unaware, given the scale of their recruiting from Apple. It’s equally hard to believe that they were aware and thought they could get away with it. I guess little from Altman would surprise me at this point. I’d like to believe that Ive wasn’t part of it, but then are we to believe that his former coworkers all kept him in the dark?

Nick Heer:

What is immediately notable about this whole thing is the myriad conflicts of interest. Apple, of course, has a current partnership with OpenAI to power Apple Intelligence features, something repeatedly mentioned on the previous version of its marketing webpage, and conspicuously absent on the current one.

But the history runs far deeper. Among the defendants in the suit is io Products — do not blame me for that capitalization choice — founded by Jony Ive and acquired by OpenAI last May. Total mentions of “Ive” in the body of this lawsuit? Zero. Also, one of the investors in Ive’s company was Emerson Collective, founded by Laurene Powell Jobs, and which now has equity in OpenAI. Neither Emerson nor Powell Jobs are defendants.

M.G. Siegler:

It all reads a bit as if Apple has been building up evidence of these allegations for some time and recently got their smoking gun. It’s not just nebulous “trade secrets” either, there are specific mentions down to “metal finishing techniques” that Apple apparently invented and protected.

It also comes a little less than a month after OpenAI was rumored to be threatening Apple with a lawsuit over the perceived failure of their deal to bake ChatGPT into the iPhone. As I wrote at the time, even just that threat was clearly not going to go over well in Cupertino.

John Gruber (Mastodon):

Footnote 13 on p. 15 states:

Apple and OpenAI have a commercial relationship involving the integration of OpenAI’s ChatGPT into Apple Intelligence. The companies have entered into a written agreement governing that integration. That agreement is not at issue here. OpenAI’s acts of trade secret misappropriation alleged herein do not arise from and have no connection to that agreement.

Be that as it may legally speaking, in practical terms it seems untenable for that Apple Intelligence partnership to continue after this.

Jason Snell:

All that bubbling-under-the-surface tension between Apple and OpenAI is now out in the open!

M.G. Siegler:

Namely, while Apple frames OpenAI – or at least its fledgling hardware division – as “rotten to its core” (interesting terminology!), they really only have a case against two and maybe three individuals. But they also clearly believe that if this case goes to trial, they will have a case against a lot more.

[…]

The parallels to the Anthony Levandowski situation after he left Google for Uber have been drawn.

[…]

It is interesting that Apple is not suing Yu-Ting “Alyssa” Peng, who is named in the suit as a co-conspirator with Liu while she was still employed at Apple, before she also subsequently went to work at OpenAI. Is it possible she’s cooperating or has cooperated with Apple in this investigation?

[…]

Even if these allegations are found to be untrue or embellished, can he reasonably stay at OpenAI now? And perhaps that’s the real outcome Apple is looking for here, as they clearly frame him as the ringleader of the 400-plus people who have come over to OpenAI from Apple.

Hartley Charlton:

OpenAI’s ambitions to build a hardware rival to the iPhone are already running into trouble because of Apple’s trade secret lawsuit, according to Bloomberg’s Mark Gurman, who argues the damage is showing up well before any court ruling.

[…]

The scale of the talent drain is a major part of why this matters to Apple. More than 400 former Apple employees now work at OpenAI, including former Apple design chief Jony Ive, and Gurman says the company poached so heavily from Apple’s iPhone product design group specifically that Apple had to rebuild parts of the team. Apple has responded with bigger retention bonuses and executives personally working to keep engineers from leaving.

Jeff Johnson:

OpenAI stole stuff? I can’t believe it! I thought they were done with stealing after they stole literally everything on the internet.

Colin Cornaby:

I’m sure there’s no problem with sending this company all your source code they pinky swear not to train on.

Steve Troughton-Smith:

I can’t believe this awful company solicited confidential information and stolen trade secrets, hiring from this other company while pretending to partner with them. ‘Rotten to the core’ indeed

😛

He’s referencing Apple doing this to Masimo. And, of course, Steve Jobs himself directed Apple to engage in illegal hiring practices.

Previously:

• OpenAI Acquires Jony Ive’s io
• Apple v. Optis Cellular Technology
• OpenAI to Become For-Profit Company
• OpenAI and Ive
• Masimo v. Apple
• OpenAI Is Today Unrecognizable
• Inside Apple’s Design Team
• BlueMail Antitrust Case Against Apple Dismissed
• Apple Suing Former A-series Chip Lead
• Apple/Google Hiring Lawsuit, Settled

Update (2026-07-17): See also: Dithering.

Tim Hardwick (FT, Hacker News):

Apple has reportedly sent legal letters to dozens of former Apple employees now working at OpenAI, telling them to preserve potentially relevant documents and communications as it continues to pursue its trade secret lawsuit against the AI company.

Update (2026-08-06): Hartley Charlton (PDF):

OpenAI has asked a federal judge to dismiss Apple’s lawsuit accusing the AI company of stealing trade secrets, calling the allegations “meritless.”

Previously:

• OpenAI Open Letter Responds to Apple Lawsuit

macOS 27 Golden Gate Beta 3 Release Notes:

Apple Intelligence for Home requires an iCloud+ subscription starting at 2TB.

Via Rajesh Pandey:

As part of the upgrade, the Home app uses AI to generate summaries of motion alerts from compatible HomeKit Secure Video cameras.

[…]

Apple likely requires the 2TB iCloud+ tier because HomeKit Secure Video already relies on iCloud+ storage tiers. The 2TB plan supports a larger number of cameras, and AI-generated video descriptions would add to Apple’s cloud processing costs.

That makes sense, but it seems odd because the iCloud tiers are named based on storage amounts, yet the feature isn’t tied to how much space you’re actually using. A lower iCloud plan with tons of free space doesn’t get the features, yet a nearly full 2 TB plan does. There are already restrictions where supporting more cameras requires higher plans, but that is at least quasi related to storage.

I reluctantly upgraded to a 2 TB family plan within the last year or so. We don’t store many files in iCloud, messages are set to auto-delete after a year, and I manually offload photos older than a year (except for hearted favorites). It’s hard to believe given how I remember the original 5 GB free tier once being sufficient, but staying under 200 GB became untenable without a lot of extra work. It doesn’t help that Live Photos sometimes gets inadvertently turned on, so that we’re storing all these little videos without realizing it.

Previously:

• “Don’t Let Me Go” and iCloud Storage Tiers
• Limiting iMessage’s Disk Usage
• The Missing iCloud Storage Bump

Rodrigo Ghedin (Hacker News):

That show of respect for its customers may change with iOS/macOS 27. Reports suggest that, at least in the first beta, Apple Intelligence is mandatory[…]

[…]

“So what’s the problem?”, you might ask. Apple’s AI takes up several gigabytes of storage and leaves less headroom for RAM.

Brandon Vigliarolo:

Those are small inconveniences, however, compared to my biggest gripe with Siri AI: It’s completely ruined Spotlight.

[…]

The new Siri-first interface that presumes that if you’re searching for anything but an app or file, you must want Siri to feed you a few links of Apple Intelligence’s choosing.

Getting to a web search from a Spotlight query now requires multiple taps: Type your query, tap “Show Results” (careful: hitting enter will trigger Siri to craft a response, eliminating the possibility of seeing any actual Spotlight content), tap on “Show More” next to the list of Siri-surfaced web results, scroll down until you see Search Google (or whatever engine you have set as your default), then tap that.

Apple Intelligence used to be opt-in, but now it seems that you can’t even opt out. I had previously mentioned the removal of the switch in the context of hypocrisy, but it seems there are enough issues here for it to warrant its own post. Also, I had previously written that the switch was combined with Siri, meaning that you could turn off Apple Intelligence if you turned off Siri, too. I now doubt that’s the case.

John Gruber:

I’m thinking that asking for a switch to turn of “Apple Intelligence” systemwide is like asking for a switch to turn off Spotlight.

Indeed, we can turn off Spotlight, both the indexing (except, alas, for APFS Time Machine volumes) and the UI, and I think that’s a good thing.

Previously:

• Apple Intelligence in appleOS 27
• No Siri AI in EU
• Chrome’s Huge weights.bin File
• Forcing Copilot AI
• Spotlight Indexing Running Wild
• Invasive Spotlight Indexing
• Bug Opted Users Back In to Sharing Siri Recordings

Apple (MacRumors):

Apple today introduced the next generation of Apple Intelligence, powered by a bold new architecture that integrates the latest Apple Foundation Models deep into Apple’s platforms and is uniquely designed to protect users’ privacy.

[…]

The next generation of Apple Intelligence also helps power Siri AI, an entirely new version of Siri.

[…]

Image Playground offers new powerful ways for users to bring their imagination to life. They can create high-quality images in virtually any style, now including photorealistic, thanks to a new generative model that runs on Private Cloud Compute. This is a major transformation for image generation across platforms. And generated images will automatically include a hidden SynthID watermark to identify them as AI-generated.

[…]

Now Messages offers one-tap suggestions based on the context of users’ conversations, making it easier than ever to get things done, such as creating a reminder or a note.

[…]

Apple Intelligence powers even more enhancements across operating systems. With automatic proofreading, users receive improved suggestions for spelling and grammar as they type across the system.

Damien Petrilli:

Apple WWDC26 “big features”: selling a rebranded gemini.

John Gruber:

What’s confusing about this Apple-Google partnership is that Google pretty much calls all things AI “Gemini”. The models are “Gemini”, the assistant is “Gemini”, and the feature integrations are “Gemini”. So Apple is taking pains to emphasize that they’re building atop the Gemini models, not the Gemini assistant.

Google creating the model, Apple figuring out how to integrate it and make it easy to use. Are we back to Apple and Google each focusing on what they do best?

Dave B.:

This is my favorite Apple event in a long time.

Instead of bombarding you with a million new features that mostly feel half-assed, this feels like it’s actually catered to the user experience instead of to marketing checklists.

Matt Ronge:

All of these AI additions to MacOS/iOS look like nice improvements, but somehow it all feels unambitious given the current state of AI.

But I suppose it’s better for them to underpromise given what happened with Siri last year…

Christina Warren:

As I expected, Apple is going to punt the “on-device” story for Apple Intelligence and push towards the “private cloud compute” story for the models that you’ll actually want to use. I’m glad on-device isn’t going away, but it’s clear a hybrid approach is absolutely necessary

Hartley Charlton:

Apple’s most advanced on-device AI model in iOS 27 requires a minimum of 12GB of unified memory, meaning the standard iPhone 17 is excluded.

Alex Rosenberg:

The 16 Pro was explicitly sold as having Apple Intelligence-capable hardware.

Juli Clover:

Apple today said it is expanding Private Cloud Compute (PCC) beyond its data centers, partnering with Google and NVIDIA to run Apple Intelligence workloads on Google Cloud.

[…]

All server components and software are part of a trusted computing base subject to verifiable transparency and no-privileged-access guarantees, plus Apple has a cryptographically verifiable ledger of all Google Cloud hardware that is part of the PCC fleet to mitigate the risk of supply chain attacks. PCC on Google Cloud also uses many of the same architectural security patterns as PCC on Apple silicon.

John Siracusa:

PSA: Remember that Apple Intelligence is disabled in macOS if you boot from an external drive.

See also: Mac Power Users Talk.

Previously:

• Apple Foundation Models in appleOS 27
• Photos AI in appleOS 27
• Agentic Password Updates
• AI-Generated Shortcuts
• Safari 27 Announced
• Siri AI Announced
• WWDC 2026 Keynote
• Gemini-Powered Siri Features Delayed
• Apple Picks Gemini
• White Label Gemini on Private Cloud Compute

Update (2026-06-19): Apple:

As we continue to refine our approach to image generation, the ImageCreator class is being discontinued and will no longer work in iOS 27, iPadOS 27, macOS 27, and visionOS 27 or later.

[…]

Transition to presenting the Image Playground sheet, which provides a consistent, system-managed image generation experience. Alternatively, you can integrate another image generation service of your choice.

Apple:

At the heart of this architecture is our third generation of Apple Foundation Models (AFM), a family of five foundation models custom-built in collaboration with Google. These span from on-device models to server-based models running on Private Cloud Compute.

Apple Foundation Models are built to unlock a wide range of helpful experiences for our users, like an entirely new Siri and intelligent tools that make everyday apps smarter and more useful.

Hartley Charlton:

“The amount of the Google Assistant we use is none,” Federighi said, explaining that Apple uses none of the Gemini models deployed to Google’s customers, none of Google’s client-side code, and no Google Search infrastructure as the knowledge backbone.

Federico Viticci:

The cloud models hosted on Private Cloud Compute are interesting: they’re not “Gemini” models; they’re Apple Foundation Models trained using proprietary data via RL, which were then “refined” with data from Google’s “frontier” models.

[…]

I’ve been covering the AFM family of models for a while now, and AFM Core Advanced is one of the most interesting on-device models I’ve read about in a while, especially in the context of model size for mobile devices and built-in multimodality with support for text, images, and audio. I’m very keen to play around with this model and understand how it holds up in practice. I wonder if the new CLI (!) for AFM will let you test this one.

Awni Hannun:

It’s very cool that Apple shipped a 20B parameter on-device.

You can’t put 20B parameters in RAM at any reasonable precision. To make it work they are using pretty exotic architecture by today’s standards.

A small model predicts from the query (or prompt) which experts to load from Nand into RAM. The key distinction from a typical MoE is that you do this once per query and then generate all the tokens with the same experts (instead of switching the experts for every token).

Greg Pierce:

Looks like there’s still a 4K context size on local FoundationModels. Pretty limiting.

Kyle Howells:

Really disappointed SwiftUI’s result builder syntax has infected FoundationModels API too.

How did a million modifier methods become the default way to build APIs?

It’s horrible for discoverability, code completion, documentation.

Steve Troughton-Smith:

FoundationModels is one of the highlights of this WWDC; I think it’s now finally what we hoped we were getting two years ago, and thought we were getting last year (just the models themselves were barely up for it then). I finally feel like I can build features with it, though I really don’t like the Sword of Damocles Apple is holding above Private Cloud Compute and how it can permanently lock you out of using the models forever across all your apps if you hit a download threshold. Poison pill

John Gruber (Mastodon):

These strict limits don’t seem to be getting as much attention as they should. It’s nice that for small developers who meet the above criteria, access to PCC has no cost. But there’s no way (yet?) to buy your way out of these limits.

[…]

The “fewer than 2 million first-time app downloads from any of their apps” restriction is particularly notable. It’s not 2 million installations for apps that are using PCC, but 2 million downloads for any app the developer has ever released.

Chad Podoski:

Is there a real time count somewhere of total lifetime downloads for a developer account, so one can plan if they at risk of the switch tripping? Reminds me of the small business qualifications. I had to try to figure out if an employers developer account was going to exceed the revenue threshold for the year. I was left trying to determine it from sales reports, not to mention sales date versus pay out date confusion. I think I botched the calculation in the end.

Anthropic (Hacker News):

Claude for Foundation Models is a Swift package that makes Claude available as a server-side language model in Apple’s Foundation Models framework. The package conforms Claude to the framework’s LanguageModel protocol, so you drive it with the same LanguageModelSession API you use for Apple’s on-device model: respond(to:), streaming, guided generation, and tool calling all work the same way.

Requests go directly from your app to the Claude API; Apple is not in the request path and does not see prompts or responses. Usage is billed to your Anthropic account at standard API pricing.

Previously:

• Apple Intelligence in appleOS 27
• Siri AI Announced
• OmniFocus 4.8.10
• OmniOutliner 6.1
• Foundation Models Framework
• Apple Intelligence Foundation Language Models

Update (2026-06-17): Simon B. Støvring:

Apple is literally stopping me from building a brand new app using Private Cloud Compute because I made a successful app a few years ago.

The new app might be a success, or it might be a complete flop. Nobody knows yet.

I would understand it if Apple were to limit developers once an app reaches a certain number of downloads or active users, but that’s not what’s happening here.

Instead, they’re blocking developers before they’ve even had a chance to build and ship the app.

Hartley Charlton (9To5Mac):

Apple today announced that the Passwords app can now automatically update weak and compromised passwords using Apple Intelligence and Safari to take action on a user’s behalf.

[…]

Apple describes the system as agentic, with Apple Intelligence and Safari securely navigating through websites, signing in, and upgrading accounts to strong passwords without the user needing to intervene beyond an initial tap.

Craig Hockenberry:

Raise your hand if you’re going to trust an AI agent with your passwords.

Maybe it’s architected so that the agent doesn’t see the actual passwords, but I find the idea of this ghost surfing through sites, logged in as me, really creepy.

Rob Mathers:

I still see Safari failing to save generated passwords in some cases.

Previously:

• Apple Intelligence in appleOS 27
• Safari 27 Announced

Update (2026-07-15): Dan Moren:

Passwords has an agentic AI feature that will scan your list of compromised passwords and then present you with a list of the ones it believes it can update for you. Even on that subset though, my experience was uneven, with just 5 of 18 updated. It’s a very cool idea, and should hopefully see improvements as the technology progresses, but right now it’s hit or miss.

Apple (MacRumors):

With Spatial Reframing, users can improve the composition of a photo after it’s been taken. Spatial Reframing builds on Apple’s deep understanding of spatial models thanks to Apple Vision Pro, so users can touch and drag a photo and preview in real time how the perspective shifts — as if they’d repositioned the camera in the original scene. Using powerful image models, Spatial Reframing will only generate new content where the perspective has been shifted, ensuring the reframed photo stays consistent with the original scene.

Users can also expand images with the Extend tool to give their subjects more breathing room. For example, they can straighten a crooked horizon without cropping out anything important, or adjust the aspect ratio, and Extend will fill in the missing pieces. Additionally, the popular Clean Up tool gets a major upgrade, so users can remove distractions with better quality and more realistic infill, even when the scene is complex.

These seem like useful features. I think it’s fine for Apple to offer them, but they do seem somewhat at odds with its previous preening about how they believe in the sanctity of photos that actually happened.

Kuba Suder:

“Reframing photos with AI” - I don’t like this at all 👎

Kind of the crosses the line to cheating about reality too much for me… it’s one thing to lighten or darken or colorize pixels, but another thing to reposition them against each other.

Louie Mantia:

“Deep respect for photography” right after a demo on generating images bc they do not respect art.

Jeff Carlson (Mastodon):

Generative AI is a technology that photographers are distancing themselves from (or should be), thanks to all the AI slop being produced everywhere. And yes, that includes creations from Apple’s Image Playground app, the image generator that the company also showed off during the WWDC keynote.

But generative AI doesn’t need to mean full images created from text prompts. When applied to selective areas, like erasing a piece of trash next to a subject’s feet, generative AI can do some of the menial work of replacing pixels that photographers would otherwise spend time retouching in an app like Photoshop. Google’s Pixel phones include a similar Magic Eraser tool.

Spatial Reframing is a great example of how the technology can be used to enhance real photos you capture.

Brian MacDuff:

I can personally confirm that Apple’s Photos “Clean Up” feature is DRASTICALLY improved in iOS 27. 😳

John Siracusa:

This is a bit hyperbolic and unkind, so I hope you can forgive me for venting. And, hey, who knows? Maybe the fancy new AI-powered reframing feature will actually get broad use. (But I doubt it…)

See also: MacRumors.

Previously:

• Apple Intelligence in appleOS 27
• Apple Intelligence in macOS 15.1 and iOS 18.1
• What Is a Photo?
• iPhone 16 Pro Camera
• Halide 2.15: Process Zero

Update (2026-06-18): Ben Lovejoy:

In the two weeks since, my view has been very much reinforced. The Clean Up tool is now significantly more powerful, while the Reframe and Extend tools address similar real-life needs …

Hartley Charlton:

Apple today announced that users can now describe a shortcut in natural language, with Apple Intelligence automatically building the automation in the background.

Marcus Mendes:

Users can simply describe what they want, such as “when I’m leaving work, text my wife with my expected ETA,” and Shortcuts will actually build the shortcut to make it work.

This sounds great in that I find creating shortcuts annoying. It’s not always clear which building blocks I should be using, and stringing them together properly seems tedious and fiddly. On the other hand, I dislike how AppleScript doesn’t get first-class support for the functionality and triggers that are available to Shortcuts.

Simon B. Støvring:

Data Jar got sherlocked, and that genuinely makes me happy.

Simon B. Støvring:

Apple should let users filter out and maybe even rewrite notifications from any app using Shortcuts.

[…]

The new notification automation trigger in Shortcuts is super cool. As far as I can tell, it won’t let me modify the notification content or filter out notifications based on the content, but it’s still a quite powerful addition.

[…]

Third-party automation triggers in Shortcuts are now possible. Sort of.

Just send a notification containing data meant for the new notification trigger to consume.

Anders Borum:

iOS 27 allows 3rd party shortcuts actions to run for longer than 30 seconds making me publish 2 actions that have been hidden behind feature flags for years.

Create cloud servers, run commands and destroy them again right from Shortcuts.

Previously:

• Apple Intelligence in appleOS 27

Update (2026-06-26): Jason Snell:

This is as close as we’ve come, across 40 years, to the original dream of putting computer power in the hands of everyone. You can literally tell your device what you want it to do, and when—”every morning show me my to-dos and calendar events for the day”—and it will generate a program to do it and a schedule to run it.

[…]

Yes, there are lots of limitations. Describe a Shortcut doesn’t work with third-party apps, only Apple’s own stuff (at least, for now). It can sometimes get confused, especially with complex queries. And it has an interesting tendency to kick things to the Use Model action—why am I not surprised that an AI model likes to build Shortcuts that themselves use AI models?

But when it works, which is most of the time, it’s magical.

Simon B. Støvring:

Shortcuts on macOS Golden Gate is… Something.

Dedicating that much window space to a larger shortcut tile and a text field feels incredibly wasteful.

Tim Hardwick:

Apple’s new version of Safari browser in macOS 27 and iOS 27 can be tasked to monitor a webpage and notify you of any changes, thanks to a new built-in feature.

[…]

In other upcoming feature additions, using the power of AI, Safari tabs that you have open can automatically be organized into topics.

Saron Yitbarek et al.:

Don’t miss our WWDC26 sessions on web technology, including What’s new in WebKit for Safari 27, to go deeper on our work in this release. Now, let’s dig into this beta, packed with 58 new features, 525 fixes and 4 deprecations that will hopefully make your work as a web developer a little easier.

Saron Yitbarek:

This year, the WebKit team is here with six sessions covering new CSS layouts, customizable form controls, 3D models, immersive spatial experiences, and browser extensions.

See also: Safari 27 Beta Release Notes.

Previously:

• WWDC 2026 Keynote
• iOS 26.2: App Marketplaces and Search Engine Choice in Japan
• How Safari Search Engine Extensions Work

Update (2026-06-17): Jeremy Keith:

But credit where credit is due. The upcoming version 27 of Safari is looking very good.

That’s not because it’s at the cutting edge of the latest web standards. Quite the opposite. Most of the changes listed for this release are bug fixes. That’s what I want to acknowledge and applaud.

Previously:

• Glow Leopard

Apple (Hacker News):

Siri AI is an entirely new version of Siri deeply integrated into iPhone, iPad, Mac, Apple Watch, and Apple Vision Pro. It can draw on personal context understanding to search across messages, emails, photos, and more, and get things done across apps with even more systemwide app actions. Additionally, Siri AI can answer questions related to the content on a user’s screen or go out to the web to get up-to-date information using broad world knowledge and generate a helpful answer. A dedicated Siri app allows users to revisit a past conversation or kick off a new one — all in one place — and uses iCloud to privately sync conversational history across a user’s products.

Apple (MacRumors, MacStories):

Powered by the next generation of Apple Intelligence, Siri AI is a completely reimagined version of Siri that is more helpful, more capable, and more intelligent. With detailed, engaging responses and natural back-and-forth conversation, Siri AI helps users get more done than ever.

[…]

Siri has been rebuilt from the ground up with powerful AI at its core. It takes full advantage of the bold new architecture for Apple Intelligence, including the next generation of Apple Foundation Models that run on device and on servers using Private Cloud Compute. When Private Cloud Compute is handling users’ requests, their personal data is not stored nor made accessible to Apple or anyone else. Outside experts can continue to verify this privacy promise at any time. Additionally, Siri AI uses the system orchestrator to tap into core capabilities like the Spotlight index and App Toolbox, which work entirely on device to keep users in control of their data.

[…]

For products that support Apple’s most advanced on-device model ever, Siri AI offers even more expressive voices, as well as a major boost in accuracy with systemwide dictation. Users have the ability to customize the expressiveness and pace of Siri’s voice so it’s just right for them. Dictation now captures what users say as polished text with greater precision, automatically handling capitalization, punctuation, and formatting as they speak. With improved speech understanding, users can speak naturally and trust that their words will appear clearly, accurately, and as intended.

I remain uninterested in these more advanced Siri features. They matter in that Apple needs to show that they’re not “behind” and that they can actually deliver on what they announce and advertise. But I don’t find the use cases compelling and wouldn’t trust the results.

If you really care about a particular concert, would you believe Siri about how the tickets work and that it will auto-create a reminder at the right time? Why should I believe that it can figure out whether a particular irregularly shaped object can fit inside an irregularly shaped container when said shapes are probably not even published on the Web? Can it really dig into my e-mail to find the right flight info? An answer without being able to see the sources is not useful. And I certainly don’t want it to pick for itself which vacation photos to send out.

I just want the basic features—from a decade or so ago, like music controls, reminders, and access to my purchases—to work quickly, reliably, and offline to the extent possible. I did not hear anything to suggest that this was a focus. Instead, everyone is celebrating how slow the demos were because that proves they weren’t fake.

Ben Thompson:

It’s actually super important that the Siri demo was kind of slow, because it emphasized it was real. Would be even better if it were live, though.

Mark Gurman:

The other interesting thing to note from the keynote: watching the demos, there is ZERO doubt these are real, recorded demonstrations on live devices. There will be no confusion about what is real or fake like there was at WWDC 2024 and the initial attempt at the new Siri.

John Gruber:

In fact, I think most of them were single takes of multiple demos back-to-back. That’s the way it should be, even when they feel a little slow.

Jonathan Wight:

Who the fuck wears a $4000 VR headset to find out if their boots will fit in their fucking backpack?

THATS the best example they could come up for that feature?

Rob Terrell:

also I wouldn’t trust the answer anyway

Kyle Howells:

These Siri demo’s feel like OpenAI demo’s from a view years ago. However, this is built in to the operating system.

Much like Dropbox vs iCloud, I feel like long term, once it’s good enough Apple will succeed in making AI a “feature” rather than a product.

Simon B. Støvring:

Apple needed to show that they could make a version of Siri that wasn’t three years behind. They didn’t succeed.

Ben Thompson (Hacker News):

And, if your standards are the state of the art in AI circa June 2024, when Apple took their first crack at answering the question, they did quite well.

[…]

What’s fascinating about this specific demo is that it also showed just how far behind Apple is.

[…]

Still, the bubble was strong enough to convince OpenAI to make the exact same mistake Dropbox did: the company somehow convinced itself that it could make enough money selling subscriptions to consumers; Anthropic, meanwhile, realized that it was enterprises who were willing to pay for AI’s massive productivity benefits, even as OpenAI failed to capitalize on their consumer market penetration by refusing to build an advertising product.

This is a long-winded way of saying that I don’t think that Apple’s agentic shortcomings are a big deal, at least for now. Agents help you do work and be more productive, and consumers don’t want to work or care about being productive. What they do want to do is watch short-form video, and an iPhone is simply much better at that than any other device ever will be; in that context, Siri being good enough is enough, and it appears that Apple crossed that bar.

John Gruber:

They need to execute, they need to prove this can scale, and most of all, they need to get third-party apps on board with App Intents and App Schemas. But it seems like they’re doing all of that. This is not a done deal but it is very realistic.

Steve Troughton-Smith:

Siri doesn’t have access to the filesystem or tool-calling, so it’s completely unable to do the kinds of tasks you might easily do with Codex or Claude. It’s a good start, but the state of the art has already dramatically shifted over the past year. The question is will Siri move in this direction, or stick to providing answers from the web?

Will Apple introduce its own distinct Codex-style agentic programming model that can do all this, and work with Xcode? You would hope so…

Aaron Pearce:

Have new Siri now. Siri still can’t do “turn lamp off and turn plug on” in one command.

Joe Rossignol:

The revamped version of Siri that is officially named “Siri AI” will be available to test in the iOS 27, iPadOS 27, macOS 27, and visionOS 27 developer betas starting today, and it is coming to the Apple Watch in a future watchOS 27 beta. “Siri AI” will also be included in the public betas of each platform that are launching in July.

“Siri AI” requires a device that is compatible with Apple Intelligence.

Hartley Charlton:

Apple senior vice president of software engineering Craig Federighi has explained why the company launched a standalone Siri app in iOS 27, after previously characterizing a dedicated chatbot as contrary to its Apple Intelligence strategy.

[…]

“We see Siri not as a separate chatbot, just an unintegrated place you go and chit-chat, but rather as an integral, conversational tool that you use in the moment, deeply integrated into your experience.”

Clear as mud.

ldt:

How to bypass the new Siri waitlist (Mac only)

See also: New Siri Waitlist.

Previously:

• No Siri AI in EU
• WWDC 2026 Keynote
• Delayed Siri Features Settlement
• Gemini-Powered Siri Features Delayed
• How Apple Fumbled Siri’s AI Makeover
• Apple Shifts Siri From Giannandrea to Rockwell
• Rotten
• Apple Delays “More Personalized Siri” Apple Intelligence Features
• Revamping Siri With iOS 18

Update (2026-06-11): famebot:

…And playlists, contacts, settings and preferences. Peel off the agentic slop bandwagon and focus on core performance, smarter caching, and resilience under adverse network conditions.

Joe Fabisevich:

Love the new Siri, sick.

Update (2026-06-12): Juli Clover has a hands-on video.

Nick Heer:

Though this is a test conducted on the very first version of Siri A.I. from a fixed point in Calgary, it seems to work quite well. So far, it is better than any version of Siri Apple has released yet and, as you can see above, it is almost as good as the original 2010 demo, before Apple acquired the company.

Update (2026-06-16): Riccardo Mori:

On the one hand, if Siri has indeed become this reliable at understanding queries like those demoed so far, it’s indeed impressive. On the other hand, I can’t help but ask myself, once again, whether people at Apple even know how regular people actually use their phones and devices.

Jeff Johnson:

Apple executives downplay and obfuscate the role of Google in Siri AI. The Siri AI announcement didn’t even mention Google or Gemini once! Does anyone actually believe Apple’s BS? The argument seems to be that since Google made a bespoke implementation for Apple, Google is not really involved, and it’s all Apple. Ridiculous.

Jeff Johnson:

It’s really odd to me that Apple is getting so much credit for Google’s work. Siri now “works” to the extent it does because Siri is to a very large extent Gemini, and Apple has wrapped that in a bow.

Apple didn’t “nail” anything. Apple failed at its own internal efforts and was forced to outsource.

This is the advantage of being a platform: you can fail and still win.

Chance Miller (via John Gruber):

Apple’s Siri team, led by Craig Federighi, held a post-WWDC keynote tech talk with members of the press this afternoon to talk through iOS 27 and the new Siri AI.

During the talk, Federighi shared more details about Apple’s collaboration with Google.

Matthew Green (Lobsters):

Apple looks like it will use some combination of Google Gemini models, combined with Google’s Confidential Inference and Apple’s own Private Cloud Compute for private hosting. These systems will process both your queries and evaluate private data from your devices.

[…]

Apple has since “expanded” PCC to encompass Google’s hardware as well. I will confess that I find the details of the new “expanded” PCC just a bit vague. It sounds a lot like Apple is primarily going to rely on Google’s existing confidential compute (running in Google datacenters) to process this data, but they’re bolting on a new layer of technical security to control which models are actually running. In any case: security experts can argue about whether this is good enough to keep Cozy Bear away from your data. What I will grant is that it’s probably good enough to keep Google and Apple from accessing your stuff, which is what most people are worried about in the first place.

So why am I so nervous?

[…]

So far we’ve discussed two adversaries: the misaligned designers of private agents (such as search operators), and the possibility of remote prompt injection attacks. But of course, in any discussion of technical privacy systems we need to talk about the last elephant in the room: your government.

Previously:

• Apple Foundation Models in appleOS 27
• Apple Intelligence in appleOS 27

Lançado junto com os iPhones 17 e 17 Pro/17 Pro Max, o iPhone Air certamente foi o modelo que roubou as atenções dessa nova safra de smartphones da Apple — mesmo não sendo lá um sucesso de vendas.

Com um design ultrafino de 5,64mm de espessura, ele traz vários recursos poderosos, como o processador A19 Pro (com uma GPU ¹Graphics processing unit, ou unidade de processamento gráfico. de 5 núcleos), uma câmera 2-em-1 Fusion de 48 megapixels, o botão de Ação (Action Button) e o Controle da Câmera (Camera Control).

Se você está atrás desse modelo, hoje o Mercado Livre está com uma promoção excelente. Trata-se da versão de 512GB, nas cores azul-céu ou branco-nuvem. Ele está saindo por R$7.499, parceláveis em até 10x no cartão de crédito!

iPhone Air de 512GB

Parcelado: de R$11.999,00 por R$7.499,00 em até 10x
À vista no Pix: de R$10.799,10 por R$6.749,10

🛍️ Produto vendido pelo Mercado Livre; aplique o cupom de desconto na página do produto💰https://t.co/7HdvQ0R2g9

— MM Ofertas (@MMOfertas) July 22, 2026

Se levarmos em conta que o preço tabelado desse modelo aqui no Brasil é de R$11.999, estamos diante de um desconto de 38%!

Para quem preferir pagar à vista no Pix, o valor fica ainda melhor: ele cai para R$6.749,10 — também um desconto de 38% em comparação com o valor cobrado pela Apple na mesma modalidade de pagamento, de R$10.799,10.

Para chegar a esses valores, aplique o cupom de desconto que aparece na página do produto.

Aproveite! 📱

Fique sempre por dentro das ofertas!

Se você não é louco, é claro que gosta de economizar. E, para isso, o MacMagazine tem diversas opções para ajudar você a conseguir comprar o seu produto Apple pagando menos!

Você pode usar a nossa extensão para os navegadores Chrome/Opera/Edge, acompanhar as ofertas pelo MM Fórum, pelo Facebook, pelo X, pelo Threads, pelo Bluesky, pelo Mastodon ou por um canal no Telegram. Escolha a melhor opção para você — ou todas, para depois não se arrepender — e economize! 😉

NOTA DE TRANSPARÊNCIA: O MacMagazine recebe uma pequena comissão sobre vendas concluídas por meio de links deste post, mas você, como consumidor, não paga nada mais pelos produtos/serviços usando os nossos links de afiliado.

Notas de rodapé

• 1
  Graphics processing unit, ou unidade de processamento gráfico.

Dando continuidade à nossa série de vídeos sobre o macOS Golden Gate 27, desta vez o foco foi em novidades que espalham inteligência artificial por diferentes cantos do sistema: Safari, Atalhos (Shortcuts), Finder, Image Playground e Fotos (Photos).

Neste segundo vídeo, mostramos o agrupamento inteligente de abas no Safari, os recursos de IA no app Atalhos e a sugestão inteligente de nomes para pastas no Finder, feita com base no conteúdo que elas guardam. Fechamos com o Image Playground e as novas ferramentas da Apple Intelligence que chegaram ao app Fotos.

Ainda tem mais pela frente nessa cobertura do macOS 27 — continue acompanhando o canal do MacMagazine para não perder os próximos vídeos! 😀

China’s cyberspace regulator has officially registered Apple’s on-device generative AI service, Apple Intelligence, for use in the country. This marks a significant step toward the long-awaited launch of the feature for Chinese consumers.

According to Reuters, “China’s cyberspace regulator said on Wednesday that Apple’s on-device generative AI service, Apple Intelligence, has been registered for use on iPhones in China, paving the way for the long-anticipated rollout of the service in the country.”

China mandates that companies register large language models and generative AI services with regulators before making them available to the public. This approval positions Apple to finally introduce its AI capabilities in one of its most important markets.

The service will integrate local AI expertise. Reuters reports, “Apple Intelligence will incorporate capabilities from AI models developed by Baidu and Alibaba, a source familiar with the matter said, speaking on condition of anonymity.”

Alibaba confirmed its involvement in a statement to Reuters: “Its Qwen model will be integrated into Apple Intelligence across Apple’s iPhone (iOS), iPad (iPadOS), Mac (macOS) and Vision Pro (visionOS) operating systems in China.”

A Baidu spokesperson also noted, “Apple is also working with Baidu to develop Apple Intelligence features for Chinese iPhone users.”

The regulator’s statement provided no specific launch date for Apple Intelligence in China.

This development comes at a crucial time for Apple in China, where the company reported a 24.4% year-on-year increase in shipments in the second quarter. The rollout of Apple Intelligence could further strengthen its competitive position in a market where consumers have eagerly awaited these AI features.

MacDailyNews Take: Clearly good news for Apple and should be for Apple’s – and Alibaba’s – stock prices, too.

---

‎
Please help support MacDailyNews — and enjoy subscriber-only articles, comments, chat, and more — by subscribing to our Substack: macdailynews.substack.com. Thank you!

Support MacDailyNews at no extra cost to you by using this link to shop at Amazon.

The post Apple Intelligence gains regulatory approval in China appeared first on MacDailyNews.

Apple is exploring new ways to keep AI processing on users’ devices, prioritizing speed, privacy, and reduced reliance on the cloud. The company is in early discussions with PrismML, a startup developing technology that dramatically compresses large AI models to fit and run efficiently on iPhones.

According to PrismML CEO Babak Hassibi, who spoke to CNBC, Apple and other companies have been evaluating the startup’s models, measuring their speed, energy efficiency, and performance on devices. “They’re really evaluating our technology right now,” Hassibi said of Apple. He described the discussions as very early but noted that “things are progressing nicely.”

Shrinking Massive Models for Everyday Devices

PrismML, a Khosla Ventures-backed spinout from the California Institute of Technology, recently released compressed versions of Alibaba’s open-source Qwen model. The company reduced the model from roughly 54 GB to less than 4 GB, enabling all 27 billion parameters to run on an iPhone 15 or newer.

The compression technique simplifies how internal information is stored—reducing values from 16 bits to just one or three possible values. PrismML claims the compressed models use 10 to 15 times less memory, generate responses six to eight times faster, and consume three to six times less energy than conventional versions.

Hassibi acknowledged a small performance trade-off: “PrismML’s models typically lose a few percentage points of overall performance, with factual recall weakening before skills such as reasoning, math and coding.” Despite this, the company is releasing two compressed versions for free, designed to run on iPhones, MacBooks, and Nvidia-powered PCs.

Boosting Siri and On-Device Capabilities

This development aligns closely with Apple’s ongoing efforts to enhance Siri. Just before PrismML’s release, Apple opened the public beta of iOS 27, introducing a major overhaul aimed at making Siri more competitive with offerings from xAI, OpenAI, and Anthropic — while emphasizing on-device processing for personal data.

Running more AI locally would bring several benefits:

• Reduced latency compared to cloud requests
• Lower cloud computing costs
• Support for features that work offline
• Stronger privacy protections, especially for sensitive health, medication, or personal information

Carolina Milanesi, president and principal analyst at Creative Strategies, told CNBC: “The more you can do on device, the better it is,” highlighting applications like computational photography, video generation, and health tools.

Analysts note Apple’s integrated hardware-software approach gives it an edge in optimizing these models. Horace Dediu of Asymco explained that Apple aims to keep most common Siri interactions on-device while routing only the most demanding tasks to the cloud. “They’re trying to figure out how big a model and how clever a model they can fit on the device,” he said.

Potential Industry Impact and Caveats

If PrismML’s technology proves effective at scale, it could influence memory requirements and datacenter demands. However, analysts caution that real-world performance — especially on lengthy prompts, multitasking, battery life, and millions of queries — remains to be fully validated.

Tarun Pathak of Counterpoint Research emphasized: “The ultimate test will be millions of queries, thousands of device combinations and robust testing at scale.” Phil Solis of IDC pointed to power consumption as a key open question for always-on or agent-like features.

Hassibi sees broader applications beyond phones, including robotics and autonomous systems where fast, local intelligence is critical: “It’s very important that the intelligence be local and that it can run fast.”

Pathak summed up the balanced future: “The combination of cloud and on-device AI can serve a more complete, efficient and privacy-centric AI experience. Complex tasks will be offloaded to the cloud, whereas sensitive, latency-critical and privacy-relevant tasks will be executed on-device.”

MacDailyNews Take: Apple’s interest in technologies like PrismML underscores its commitment to privacy as a core differentiator while pushing the boundaries of what smartphones can achieve with AI. As testing continues, this could mark a significant step toward more capable, personal, and private intelligent assistants.

---

‎
Please help support MacDailyNews — and enjoy subscriber-only articles, comments, chat, and more — by subscribing to our Substack: macdailynews.substack.com. Thank you!

Support MacDailyNews at no extra cost to you by using this link to shop at Amazon.

The post Privacy-focused Apple in talks with startup to run more powerful AI models directly on devices appeared first on MacDailyNews.

While many Big Tech peers chase high-risk AI growth, Apple stands out for its stability and resilience. The company is projected to grow free cash flow in 2026 — a rarity among its counterparts — positioning its stock as a defensive haven traditionally reserved for consumer staples, utilities, and healthcare names.

Nate Wolf for Barron’s:
‎

“No one admits they like or want to buy more AAPL,” Mizuho Securities analyst Jordan Klein wrote Friday. “But I get why more folks [are] looking at AAPL as a hedge or long idea.”

“Analysts estimate Meta Platforms, Oracle, and Amazon will have negative free cash flow in 2026 because of their AI investments. Meanwhile, Apple is expected to produce $143 billion in free cash flow, up from $109 billion last year, per FactSet. Its capital expenditures will likely decline in 2026.

“[Apple] has definitely been out of favor, but carries a lot less direct near-term execution risk,” Klein wrote.”

But Apple is one of the few high-profile tech names that isn’t a primary consumer or supplier of AI compute, and that differentiation has value… If Apple’s share price is any indication, plenty of investors want to keep at least some money away from the AI trade.

‎
MacDailyNews Take: As we wrote back on February 18 when Apple shares closed at $264.11 ($50 per share ago):

Just wait until Apple begins to release Apple Intelligence features underpinned with — but privacy-protected from — Google Gemini and things like Siri begin to actually work as well as, or even better than, people expect them to work in 2026. The sky’s the limit for Apple stock!

---

‎
Please help support MacDailyNews — and enjoy subscriber-only articles, comments, chat, and more — by subscribing to our Substack: macdailynews.substack.com. Thank you!

Support MacDailyNews at no extra cost to you by using this link to shop at Amazon.

The post Apple emerges as the ultimate defensive play in an AI-driven market appeared first on MacDailyNews.

Apple is in discussions with PrismML, a Khosla Ventures-backed startup that claims to have achieved a major breakthrough in running exceptionally large AI models directly on iPhones without relying on cloud servers, Aaron Tilley reports for The Information. The company says it has successfully compressed and executed a 27-billion-parameter model on the latest iPhone hardware, marking what it describes as the largest AI model ever run fully on-device.

PrismML, which emerged from stealth earlier this year as a Caltech spinout, specializes in extreme model compression techniques, particularly 1-bit and ternary weight architectures. These methods drastically reduce memory footprints while aiming to preserve performance. According to the startup, its technology allowed Alibaba’s open-source Qwen 3.6 (27B parameters) — which normally requires about 54 GB in standard precision — to shrink to under 4 GB. The compressed model reportedly ran with all parameters active simultaneously on an iPhone 17 Pro, enabling more complex on-device tasks like advanced reasoning and software development assistance.

This development aligns with Apple’s long-standing push to enhance Apple Intelligence with more powerful on-device capabilities. By keeping AI processing local, Apple can improve privacy, reduce latency, and lessen dependence on cloud infrastructure and external partners. Sources familiar with the matter indicate Apple has held talks with PrismML about potential integration of its compression technology to expand what future iPhones can handle natively.

The Technical Leap: 1-Bit Models and Intelligence Density

Traditional large language models use 16-bit floating-point precision, leading to massive memory requirements. PrismML’s approach uses 1-bit weights (essentially on/off states) and ternary variants (weights of -1, 0, or +1). The company’s flagship open-source “Bonsai” family includes:

• Bonsai 8B: An 8-billion-parameter model that fits in roughly 1 GB (versus ~16 GB for a typical 16-bit equivalent), while delivering competitive results on benchmarks like MMLU, GSM8K, and HumanEval.

• Smaller variants (4B and 1.7B) optimized for even greater efficiency.

PrismML claims these models offer up to 14× smaller memory use, 8× faster inference, and 5× better energy efficiency compared to full-precision counterparts, resulting in over 10× “intelligence density.” The startup has open-sourced its models under Apache 2.0 and released custom kernels for Apple’s Metal framework, making them compatible with iPhone and Mac hardware.

Co-founders, including Caltech professor Babak Hassibi and other PhDs, emphasize that the breakthrough stems from years of research into neural network compression. Investors include Khosla Ventures, Cerberus Capital, and Caltech itself, with the company raising $16.25 million in seed funding. Vinod Khosla has praised the work as a “mathematical breakthrough” that could shift AI from data-center dominance to efficient edge deployment.

Implications for Apple and the Industry

For Apple, adopting such technology could supercharge features in iOS and Apple Intelligence, enabling more sophisticated AI agents, real-time coding help, or advanced multimodal capabilities without constant server calls. It also addresses growing concerns around energy consumption and data privacy in AI.

PrismML positions its tech not just for consumer devices but for robotics, wearables, and industrial edge applications. By open-sourcing core models, the company hopes to accelerate adoption and invite community improvements.

While independent verification of all performance claims is still emerging, the reported iPhone demo and Apple’s interest signal strong validation. As the AI race intensifies, solutions that pack more intelligence into everyday devices — without ballooning costs or power draw — could prove transformative.

PrismML plans further releases and is expected to continue engaging with major device makers.

MacDailyNews Take: For Apple, which has invested heavily in on-device AI silicon like the Neural Engine, this could represent a key for the next-gen of truly personal, secure, and private artificial intelligence.

---

‎
Please help support MacDailyNews — and enjoy subscriber-only articles, comments, chat, and more — by subscribing to our Substack: macdailynews.substack.com. Thank you!

Support MacDailyNews at no extra cost to you by using this link to shop at Amazon.

The post Apple eyes PrismML to run huge AI models directly on iPhone appeared first on MacDailyNews.

Enables direct image editing in Pixelmator Pro for Apple Creator Studio subscribers. ($129.99 annual Apple Creator Studio subscription, various sizes, macOS 15.6+)

O Apple Creator Studio desbloqueia vários recursos bacanas de aplicativos já conhecidos pelos usuários de dispositivos da Maçã.

Dentre os exemplos, estão os apps Pages, Numbers e Keynote. Com eles, assinantes do Creator Studio podem gerar imagens e adicioná-las aos seus documentos, planilhas e apresentações de slides.

Veja como fazer isso nos parágrafos a seguir! ✨

Como gerar imagens usando o Pages, Numbers ou Keynote pelo iPhone/iPad

1. Com um dos apps abertos, abra o arquivo onde deseja anexar a imagem.
2. Escolha o ícone roxo (na barra superior) e vá em “Gerar Imagem”.
3. Selecione o estilo, a forma de visualização e a proporção desejadas.
4. Descreva a imagem no campo da parte inferior da tela.

Após ela ser gerada, você pode fazer outras edições e salvá-la, por exemplo.

Como gerar imagens usando o Pages, Numbers ou Keynote pelo Mac

1. Com o arquivo aberto, clique no ícone de explorar biblioteca (na barra superior) — ou, na barra de menus, vá até Inserir » Gerar Imagem.

2. Em seguida, escolha “Gerar Imagem”.

3. Siga os passos a seguir para gerar, aprimorar e anexar a imagem desejada.