Eshu Marneedi

Computer Science Was Never About Software Engineering

Eshu Marneedi — Sat, 17 Oct 2026 04:45:17 -0400

Nathan Wilmers, Regina Catipon, Maxim Massenkoff, and Peter McCrory, in an Anthropic research article:

How has AI restructured work and employment at Anthropic? As a company at the frontier of AI adoption, Anthropic could be a leading indicator of how AI will shift work organization, hiring, and labor demand across the economy.

Anthropic has been hiring for a smaller share of junior technical positions. This occurred as hours worked by AI agents relative to employees rapidly expanded. By August interactive AI agent hours outnumbered employees’ 6 to 1.

As AI agents diffused through the company, Anthropic began producing more code per person, closing more sales per rep, and processing more contracts. Following the same employee over time, output rose as they adopted new AI tools.

This acceleration makes human attention the bottleneck for agent output. Even among non-technical employees, reading all the output of their Claude agents would take half their work week. In response, employees turn to “meta-work,” of building visibility tools and monitoring agent pipelines.

Changes in work organization are reflected in shifting priorities among hiring managers, who increasingly emphasize autonomy and judgment, relative to technical domain expertise and implementation…

Anthropic has responded to these changes by setting up an early warning system to track jobs at risk of disruption, and as jobs evolve the company has prioritized internal redeployments. These actions, taken at the frontier of both AI development and workplace change, show how specific organizational commitments can shape the impact of AI on employment. But they also raise difficult questions about exactly what those commitments should be.

Anthropic has not “responded” to these changes as much as it has caused them itself. Not by developing Claude, but by intentionally reducing hiring as a result of greater emphasis on “autonomy and judgment.” Setting up an early warning system for jobs being intentionally disrupted is cyclical in some sinister way.

But the emphasis on “autonomy and judgment” is not unwarranted, nor inherently negative — it just doesn’t need to come at the cost of entry-level jobs, which are fundamental to a functioning economy. Junior-level interns and full-time employees can, too, exercise judgment and take advantage of autonomy, and giving them the opportunity to do so at one of the most influential technology companies of our generation is a great way to positively influence artificial intelligence’s economic impact. What I’m saying here is that if we never allow entry-level workers to hone and exercise their judgment, relegating them to lower-level jobs that are, by Anthropic’s own admission, fading, we’ll have a concerning deficit of senior-level skilled workers in a few decades.

There’s a lot to like about the current trends in tech hiring. “Technical” skills — or, more specifically, the credentials to prove those skills — have never been fairly distributed among laborers. There are many more talented computer scientists than those who come out of Stanford University, many of whom lack degrees altogether. Technical interviews, I’d argue, are moot in the age of automation. As AI reshapes hiring, I hope to see emphasis on university prestige and technical interviews fade into obscurity. But this does not mean that we shouldn’t be hiring entry-level employees, nor that there isn’t a way to measure their judgment and capabilities without formal credentials. If the emphasis is on autonomy and judgment, hire free thinkers and people with great ideas. Test them on those ideas, not how well they’ve memorized the syntax of C++. If the latter is what recruiting is optimized for, then yes, there’s no point in hiring an entry-level computer science major.

But computer science is not a vocational degree, and candidly the only people who think it is are venture capitalists who studied it for the money. To them, I say, Get Out of Technology. Computer science comprises a unique blend of the liberal arts, humanities, and engineering. There’s a reason computer science isn’t called “information technology,” and why most computer science courses hinge so strongly on theory, not application. Computer science students don’t “learn to code for four years,” as one ill-informed internet commenter I found said. Most of them — the ones who are least likely to drop out — have already learned and built software by themselves. To gain admission to a relatively prestigious university’s undergraduate computer science program, it is almost a prerequisite that you’ve already become a software engineer by the time you’re 16.

It’s time to start treating computer scientists like philosophers and drop the act of pretending computer science was ever about software engineering. When autonomy and judgment are paramount to getting a job at one of the most influential AI companies, computer science students are the most natural fit. They learn over four years to not only engineer software, but think deeply about computation and automation itself. Good computer scientists know not just how to automate, but what should be automated at all. Learning programming languages and solving hundreds of LeetCode problems — as the research paper posits — does not teach this. Test computer scientists on the judgment they should have spent four years learning about and refining. New computer scientists are not entry-level software engineers, and when we stop treating them as such — or expecting them to assume the title of software engineer — suddenly their education carries much more value.

We should absolutely be hiring entry-level computer science majors to think through the toughest questions about the philosophy of computing, automation, and data. We should be using their four years of experience to shape AI development, not to engineer software. And once the industry catches up to this — I think it will eventually — it’ll prove that there has never been a better time to study computer science than during an era where computers are poised to usher in a new societal order.

Apple Introduces HomeView, Updates Apple TV 4K and HomePod mini with Siri AI

Eshu Marneedi — Wed, 14 Oct 2026 04:17:45 -0400

Apple Newsroom, during Tuesday’s “Welcome Home” press briefing in New York:

Apple today introduced Apple HomeView, designed to be central to the home and personal to everyone in it. With a stunning 7-inch Liquid Retina display, HomeView brings entertainment, connection, and smart home management together in one place, from playing music and making FaceTime calls to home monitoring and hands-free automations with voice control — all designed with privacy and security at the core. HomeView recognizes users by their faces, adapting so they can access their personal content in Messages, FaceTime, Calendar, Reminders, and Find My. HomeView features Siri AI, an entirely new version of Siri that’s profoundly more capable and conversational, with broad world knowledge that powers richer music discovery and onscreen awareness. When paired with an Apple Intelligence-enabled device, it can also draw on a user’s personal context, such as their Calendar, Messages, and relevant information across their apps, to provide more helpful, personalized responses.

Apple HomeView — I’m still debating whether or not I put a “the” in front — runs a special unnamed operating system that runs only on the device. It seems to be a weird mélange between watchOS and tvOS, with stripped-down versions of the aforementioned Apple apps. The most interesting note about the product, to me at least, is that it doesn’t directly support third-party apps: Rather, it runs certain audio and video apps, like Spotify and Audible, through, peculiarly, CarPlay. Developers do not write new apps or optimize their existing apps for HomeView at all; if they make an audio or video app for CarPlay, it will be available to use on HomeView. These apps don’t run on HomeView directly, though: Much like CarPlay, connected iPhones run the apps while HomeView merely broadcasts the user interface. (This is from excellent reporting by Joanna Stern over at The New Things.)

Siri AI is the centerpiece of HomeView, and rightfully so. I continue to think Siri AI is one of the most compelling artificial intelligence products of the era, and it should be brought to the home. When Siri AI taps into the personal context, it either recognizes a user by voice — like all HomePods — or through camera-based facial recognition. This is intentionally and carefully not called Face ID because it does not use the almost bulletproof facial recognition technology on iPhones and iPads. I suppose it can be tricked by photos or identical twins in ways Face ID cannot. But it’s presumably accurate enough to secure text messages, calendar events, and more. Siri AI in iOS 27 also benefits from visual responses, and having those on a screen will be particularly helpful. Again, Siri AI doesn’t run on HomeView itself; it runs on connected iPhones. This means that if your iPhone doesn’t support Siri AI, buying a HomeView won’t let you use it, either. (It also means that HomeView is not for sale in the European Union.)

HomeView is a compelling product for anyone who wants a HomePod with a screen. At $300, I find it reasonably priced, especially now that the HomePod mini is $150. That market is narrow, but it must exist somewhere, and HomeView sells at a good price for those buyers. For everyone else, I find the omission of third-party apps unacceptable. If the idea is for HomeView to be an iPad for the home, it fails spectacularly. I don’t think games or work-intensive apps are fair game for HomeView, but certain recipe books, browsers, note-taking apps, and to-do lists ought to be on this device. People use calendaring services other than the stock Calendar app. Many families have grocery lists and other tasks in third-party apps like AnyList and Due, not Reminders. In fact, Reminders is relatively ill-suited for families — it’s almost too simple. And what about third-party messaging apps, like Discord or Instagram, which are used by teenagers more than Messages?

So HomeView is not an iPad. It’s a home-control device first and foremost, displaying ambient information like the weather, time, widgets, or Now Playing. Again, I find that framing perfectly acceptable — it’s a HomePod with a screen; the HomePod does not support third-party apps, or any apps for that matter — but Apple itself seems confused about what the product is intended to be. Why does it display personal information like messages or reminders at all if it’s clearly ill-suited for those duties? Why does it have a web browser if the only way to open it is by using iPhone Handoff to transfer a webpage from iOS to HomeView? (I will admit the feature that cleans up recipes is nice; that should come to iOS.) I will frame my question more explicitly: Is this product an iPad or a HomePod? If it’s a HomePod, it’s doing too much, and if it’s an iPad, it’s doing too little. HomeView is a fundamentally confused product, marketed in a product launch video that frankly is one of the worst Apple advertisements in modern history. Much like the characters in that video, HomeView tries too hard and fails too much.

Also from Tuesday:

Apple today announced the next generation of Apple TV 4K, delivering its most immersive home theater experience yet. The new Apple TV 4K features Siri AI, an entirely new version of Siri powered by the next generation of Apple Intelligence, transforming how users search for, discover, and enjoy their favorite movies and TV shows through a full-screen Siri AI experience. With built-in far-field microphones, Apple TV 4K also lets users talk to Siri completely hands-free.

Apple really buried the lede in a footnote here: Siri AI is available in beta for all Apple TV 4K models, second-generation and later. The reason why is because, again, Siri AI runs on connected iPhones, not on these products themselves. So the new Apple TV 4K adds two new features: surround sound using HomePods, and a microphone to speak to Siri hands-free. (The Apple TV 4K does not have a speaker, so it’s not a HomePod replacement. Rather, the microphone just means you don’t have to hold up the Siri Remote to speak to Siri.) The refreshed Siri Remote is available standalone for $60 and includes a speaker and better Precision Finding capability, which seems like it’s worth the price. The refreshed Apple TV 4K also moves the indicator light to the center and includes an A19 processor with Wi-Fi 7.

For this — surround sound with HomePods and a microphone for Siri — Apple asks $300, a 2.3-times increase in price over a few months ago, when the Apple TV used to cost $130. Again, for a 2.3-times increase in price, the only new features are surround sound with HomePod mini and a microphone. There are no sensible words to describe how ludicrously overpriced the new Apple TV 4K is. I must imagine the margins on this product have always been sky-high, and Apple is intent on maintaining them throughout the memory shortage. This product doesn’t even have that much memory! Yes, the Apple TV 4K is leagues ahead of every one of its competitors, and if I didn’t own one today, I’d probably pay $300 to get one. I own three myself and love them. But that is a pathetic excuse to charge $300 for a product whose software receives almost no attention from Apple year over year. This is asinine.

Speaking of unacceptable:

Apple today announced HomePod mini 2, featuring Siri AI, gorgeous new colors, and enhanced audio experiences. Siri AI, coming to HomePod mini 2 in a future software update, is a profoundly more capable and conversational assistant with personal context understanding, broad world knowledge that powers tasks like richer music discovery, and the ability to tap into even more apps — bringing a whole new level of capability to HomePod mini.

The new HomePod mini¹ at least adds Siri AI and a new feature called Sound Glow, where the top display shows the colors of the music’s album artwork. Still, it retails for $150, a $50 price increase for a product that has almost no memory. It’s running a last-generation Apple Watch processor — in what world does that cost $50 more? I am not one to complain about Apple product prices. I gladly pay top dollar for the highest-quality Apple products because I know they’re reliable and useful. The Apple TV has always been more expensive than any of its competitors, but it’s so ahead of them that it feels unfair to even call those products its competitors. I’m not of the opinion that the “Apple tax” is a genuine phenomenon, but I’m strongly reconsidering my opinion. The HomePod mini is not worth $150, and Apple missed the mark on all three of its product launches on Tuesday.

The big HomePod once again lives on only in our hearts. ↩︎

An Abridged iPhone 18 Pro Review

Eshu Marneedi — Mon, 12 Oct 2026 08:35:40 -0400

Brevity comes easy this year

iPhone 18 Pro.

A few weeks after this article’s publication, a much more exciting iPhone will go on sale: iPhone Duo, the hotly anticipated foldable iPhone that leaves room for many more words and photographs. But its flagship sibling, iPhone 18 Pro, deserves at least a few notes. There’s not much to discuss here, but the little there is makes iPhone 18 Pro the default iPhone to buy heading into the busy holiday season.

I’m keeping this review short — much shorter than my annual iPhone reviews typically are — because, again, there’s just so little ground to cover. Frequent readers know my usual spiel about how each iPhone generation is a refinement and perfection of the last, and this year’s model is no different, honing the tried-and-true iPhone design. It reduces the size of the now four-year-old Dynamic Island, adds new Photographic Styles, introduces a variable-aperture main camera, and improves battery life and performance. It’s easy to discount these updates, but they truly make a difference in day-to-day use.

So that’s what I’ll focus on: performance, Photographic Styles and the camera, and the Dynamic Island. And I promise it’ll be succinct. Brevity comes easy this year. John Gruber said it right at Daring Fireball: iPhone 18 Pro is boring, but in a very good way.

Performance

iPhone 18 Pro runs noticeably cooler.

I haven’t written about an iPhone’s performance in five years; the last time was when I reviewed iPhone 13 Pro in 2021. Modern iPhones are so performant that there’s almost nothing to remark on from year to year, but this year’s A20 Pro system-on-a-chip is remarkable — so remarkable that it unquestionably deserves space in this short review. The A20 Pro is Apple’s first semiconductor design built on Taiwan Semiconductor Manufacturing Company’s 2-nanometer process, marking three years since the last process transition. I was wary of this in my event impressions article from September, since the A17 Pro and M3 processors infamously suffered from growing (shrinking?) pains, namely thermal throttling and poor battery life.

I’m delighted to report that the A20 Pro is an leap over its predecessor in every dimension. App launches are noticeably faster than on my last-generation iPhone 17 Pro, even on the iOS 27 beta. The new processor also yields significantly better battery life generation over generation, and I’ve been unable to deplete the phone’s battery even after a day and a half of heavy use. Charging speeds are also markedly faster with a high-wattage power adapter: Using the 96-watt one that came with my MacBook Pro, I was able to charge the device from 0 to 80 percent in about 45 minutes, and from 0 to 50 percent in less than 20 minutes. Fast charging is such a useful quality-of-life feature, and Android flagships have had it for years.

Perhaps the most noticeable improvement — and the most heavily marketed — is the expanded vapor chamber cooling assembly inside the device. My iPhone 17 Pro, in contrast to its titanium predecessors, would not get warm unless it was either charging wirelessly or running the camera for a while. iPhone 18 Pro never overheats, even with the camera in use. It still gets warm when, say, filming 4K video, but it never throttles or becomes too uncomfortable to hold, even while fast charging. TSMC’s 2-nm process is rock-solid, and the A20 Pro combined with the improved cooler makes for a welcome upgrade. Anyone with a 3-nm iPhone will instantly notice the difference from the moment iOS restores from a backup. (That has always been my litmus test for whether an iPhone is thermally compromised.)

Photographic Styles 3 and the Variable-Aperture Main Camera

iPhone 18 Pro’s cameras are improved over last year.

The standout new hardware feature of this year’s iPhone generation is the variable-aperture main camera lens. Aperture refers to how much light the camera lets in, measured in f-stops: A lower number, like 𝑓/1.8, indicates that more light hits the sensor, while a higher number, like 𝑓/4.0, means the lens is more constricted. Aperture is one of the main levers of photography, alongside shutter speed, ISO, and focal length. It also determines the depth of field in an image, and with it, the amount of bokeh. A lower f-number yields a blurrier background, known as a shallow depth of field.

Most smartphone camera lenses, including those on prior iPhones, have a fixed aperture, usually somewhere in the middle to accommodate both bright and dark scenes. Aperture is a physical attribute of the lens not controlled by software, and so iPhone 17 Pro and iPhones before it had a fixed main camera aperture of 𝑓/1.78. iPhone 18 Pro introduces a variable-aperture camera lens through small blades arranged in a circle in front of the camera sensor. These blades can be expanded or contracted electronically with electromagnets to let in more or less light. The result is that iPhone 18 Pro’s main camera can shoot at four apertures, selectable in the Camera app: 𝑓/1.48, 𝑓/1.8, 𝑓/2.8, and 𝑓/4.0.

iOS defaults to 𝑓/1.8 in most scenarios, mimicking the hardware-defined aperture from years prior. That aperture was chosen for its versatility across a variety of scenes, and it works well most of the time. In low-light conditions, iOS automatically chooses 𝑓/1.48 and shortens the Night Mode exposure time to just a second or two. This accomplishes two things: (a) iOS no longer requires people to hold their phones steady for as long when taking a photo in the dark, and (b) the resulting photos look more natural, since more light physically reaches the sensor and the camera relies less on ISO. Night Mode stitches multiple photos of different exposures together to produce a brighter image. Standalone cameras adjust exposure using a combination of shutter speed, ISO, and aperture, and iPhone 18 Pro lets iOS modify that third variable, making Night Mode more effective.

The results are subtle but stunning. iPhone 18 Pro captures low-light shots almost as well as standalone mirrorless cameras with large sensors. In dark scenes, the muddiness and noise that came with software processing and artificial light are reduced, and lights are more vibrant. In more moderate lighting, colors appear livelier and almost more saturated, while shadows are more luminous. In bright environments, like a midday landscape, grass and trees appear brighter and sharper. Again, the differences between generations are tenuous, but the added flexibility when shooting makes iPhone 18 Pro that much better a day-to-day camera. I’ve captured some genuinely gorgeous shots, both indoors and outdoors.

Left: iPhone 18 Pro at 𝑓/2.8. Right: iPhone 17 Pro at 𝑓/1.78.

Left: iPhone 18 Pro at 𝑓/1.48. Right: iPhone 17 Pro at 𝑓/1.78.

Photographic Styles have been one of my favorite iPhone camera features in recent years, and this year’s updates, in tandem with the variable-aperture lens assembly, make them even more compelling and more fun to shoot with. The new Photographic Styles add a separate tab for adjusting texture, alongside the Tone, Color, and Palette controls introduced two years ago. The three options are Soft Skin, Glow, and Film, and Film seems the most worth using. (Soft Skin and Glow seem like they’ll be popular on Instagram.) For each effect, you can tune the intensity and add or remove film-like grain. (You cannot combine effects.) The Film effect adds grain automatically, and its intensity slider controls how much. Even with grain turned off, selecting Film tweaks the image’s colors to look more artisanal.

Dramatic + Film, taken on iPhone 18 Pro at 𝑓/1.48.

Many of the old preset Photographic Styles now include the new effects; for instance, Amber is now Amber + Glow. It’s possible to remove these effects if you don’t want them — the “undertones,” as the color selections are called, are independent of the effects. Modifying a Photographic Style saves it for later use. As I said in my iPhone 16 Pro review, I think the editor for modifying Photographic Styles makes a lot more sense on the Mac, where the various selections are laid out more logically. There are 16 undertones to choose from — effectively color profiles made by Apple. From there, you can adjust the Tone, Color, and Palette, or add the Soft Skin, Glow, or Film effects. On iOS, all of these controls are buried behind various unintuitive context menus, and as Photographic Styles grow more complex, I think this interface needs an overhaul.

Cozy + Film, taken on iPhone 18 Pro.

The new Photographic Styles are a joy to shoot with, but I find it odd that they’re unavailable on older iPhones. My guess is that the computational image pipeline has been reworked such that only iPhones with more Neural Engine cores can shoot photos compatible with Photographic Styles 3, but I don’t fault anyone who sees this as an intentional cash grab. The new Photographic Styles can be edited on an older iPhone or Mac running iOS 27 or macOS 27 Golden Gate, but the photo must have been taken with a new iPhone.

Rose Gold + Film, taken on iPhone 18 Pro.

Bright + Film, taken on iPhone 18 Pro.

The new Camera app is also needlessly confusing and hides this year’s most prominent feature behind yet another context menu. The main Photo tab indeed has an f-stop icon (𝑓) to the left of the shutter button, but this only toggles the artificial Portrait Mode, which, as I’ve said in many past iPhone reviews, is unnecessary. iPhone 18 Pro has an even shallower depth of field thanks to the 𝑓/1.48 aperture, further undercutting Portrait Mode, so I just don’t see why it deserves such a conspicuous spot. The Camera app in iOS 27 does allow some toolbar customization, but it’s impossible to swap this Portrait button for something else. For now, I’ve customized the toolbar to drop Live Photos — who uses those? — and add manual aperture control. For all other customization, I use Halide, which recently gained manual aperture control, too.

iPhone cameras have progressed so far that hardly anyone carries around a point-and-shoot anymore. That doesn’t mean I don’t own a proper mirrorless camera — I used it to take the photos for this review — but I reach for it far less than I did even six years ago. This year is no exception to the steady stream of camera improvements Pro-model iPhones have enjoyed, and I think anyone upgrading will appreciate the latest batch.

Dynamic Island

iPhone 18 Pro’s Dynamic Island is now smaller.

The Dynamic Island has become one of the most beloved iPhone features in recent memory, so much so that I’ve never once complained about its size, nor heard anyone else comment on it. This stands in direct contrast to the notch from iPhone X to iPhone 13, which was briefly shrunk only to be foregone entirely for the Dynamic Island. I dare say my iPhone would feel broken without the Dynamic Island, for I find it so useful in the littlest ways. It’s nice to have live sports scores, Now Playing, or other at-a-glance information pinned to the top of the screen at all times.

This year, the Dynamic Island — by virtue of its reduced horizontal footprint — can display three Live Activities simultaneously, with an asterisk: Two of those Live Activities can be from third-party apps, but at least one must be from a first-party service, like Now Playing or Screen Recording. I’m perplexed by this limitation, because the third Live Activity space is no smaller than the minimal appearance third-party apps were forced into when two Live Activities ran on pre-iPhone 18 models. On iPhone 18 Pro, the size classes now work as follows:

One Live Activity: apps can use the full status bar — both the leading and trailing sides — to display content.
Two Live Activities: apps can take only one side, but the Live Activity needn’t be minimal, as described in Apple’s Human Interface Guidelines. (In prior generations, apps were forced into their minimal appearance.)

Two Live Activities.

Three Live Activities: two apps and one system Live Activity are all displayed in their minimal appearances. (This was not possible in prior generations.)

Three Live Activities.

One interesting side note is the placement of the infrared camera on iPhone 18 Pro: It is now under the display and positioned to the left of the Dynamic Island. It’s nearly impossible to tell from photos, but the pixel density in that small circle directly over the time in the status bar is lower than that of the rest of the display. Internet sleuths have made this out to be a much bigger issue than it is — it’s hardly noticeable in typical use — but if you peer closely, it’s definitely visible. I think the bigger Live Activities make up for the small loss in pixel density over the clock’s minute digits.

The infrared camera is invisible from nearly every angle.

Miscellany

The C2 cellular modem, available in all international models and U.S. iPhones 18 Pro (not Max), is just about as reliable as the Qualcomm modem before it. I don’t notice any more or fewer bars of Verizon service in my suburban neighborhood, and speeds are roughly the same on 5G C-band; there’s nothing to worry about here. The transition to Apple modems was always about efficiency, not performance, and all C-series modems have done exceptionally well, even making iPhone Air last all day with 5G. Word on the street is that Qualcomm’s contract with Apple stipulates that, until the contract expires, at least one iPhone model must use a Qualcomm modem, and the U.S. iPhone 18 Pro Max is the result of that lawyering.
This is the first year I didn’t buy the “flagship” Pro iPhone color, Burgundy. I have no regrets — the Glacier Blue iPhone is gorgeous. It’s much deeper and more metallic than Sierra Blue from iPhone 13 Pro, and it looks different in every lighting condition — though nothing like the product photos. I’d be a liar if I said it weren’t underwhelming next to last year’s Cosmic Orange bombshell — I’d be astonished if Apple didn’t bring that shade back in a future year — but I’d still recommend Glacier Blue over any of the other colors.

iPhone 18 Pro in Glacier Blue.

iPhone 18 Pro is undoubtedly the flagship iPhone of the year, just as much as any S-model iPhone would have been back when Apple still used the moniker. Even — I’d even say especially — considering the looming iPhone Duo, it’s still the best iPhone, with the best camera, screen, and battery life. If there’s anything to take away from this review, it’s that iPhone 18 Pro is not an iPhone 8 situation, i.e., the last of its kind. iPhone Duo is obviously the more exciting news of the season, but for everyone else, there’s the classic iPhone, and it’s never been better.

I told you I’d be brief.

LG Plans to Launch HomeKit-Enabled Accessories, With Apple’s Blessing

Eshu Marneedi — Sat, 10 Oct 2026 04:04:29 -0400

Mark Gurman, reporting for Bloomberg:

Apple Inc.’s upcoming push into smart home devices will include a doorbell, thermostat and other accessories developed through an unusual partnership with LG Electronics Inc.

The products will be part of an ecosystem of devices that work with Apple’s new smart home hub, as well as an upgraded HomePod mini and fresh TV set-top box, according to people with knowledge of the matter. The new Apple devices are set to be introduced on Oct. 13.

The products developed with South Korea’s LG will also include a smart deadbolt lock, indoor security camera, outdoor security camera, and floodlight camera, said the people, who asked not to be identified because the undertaking is private.

It is lazy reporting to call this partnership unusual. Back in the dark days of the Mac, when Apple discontinued its only first-party display, it replaced it with the LG UltraFine 4K and 5K monitors: the LG-made iMac LCD panels packaged as standalone monitors. Most third-party monitors either come in the wrong pixel density (e.g., 27-inch 4K) or lack software support for the Mac. The LG UltraFine monitors, while not in Apple’s chassis, worked just like Apple monitors, with not even as much as discrete volume, brightness, and power controls. It’s just that Apple, which at that point had all but given up on the Mac, couldn’t invest the resources into building a first-class monitor for it, so it contracted the job to LG. (Apple eventually took the job back in 2022 with the launch of the overpriced Studio Display.)

All of this is to say that whenever Apple doesn’t care enough about a market but needs a product in a given category, it contracts it to a well-connected, trustworthy third-party supplier. LG made the display panels for the iMac, so it was natural that it would make standalone displays for other Macs, too. Another pertinent example is Belkin, which makes many iPhone chargers and cables sold at Apple stores and recommended by Apple employees; it made the only HDMI cable designed for the Apple TV 4K when it was announced. Belkin is owned by Foxconn, the company that manufactures iPhones — it has deep ties to Apple, and when Apple doesn’t want to make some charger or cable, Belkin does. One could be fooled into thinking that Belkin is a subsidiary of Apple.

I’ve never seen Apple as a smart home company. The HomePod was quite literally canceled after the HomePod mini before it was resurrected in 2023; the Apple TV goes years without updates; tvOS is not even mentioned in some Worldwide Developers Conference keynotes; and the AirPort was far too short-lived. I cannot remember the last time the Home app received an update — it doesn’t even support the latest Matter devices, like smart vacuums. So it does not come as a surprise that Apple, on the cusp of announcing a new smart home device, wants to have new accessories to market alongside it — but it doesn’t care enough to manufacture them itself. There’s an argument that Apple may want to make products like these by itself in the future, but that has to be far away. Apple is not a smart home company, and smart home appliances don’t sell as well as Apple’s other products. This is a good middle ground.

In particular, the HomeKit Secure Video-enabled security camera market is bleak. Most people who care about HKSV buy other cameras and hook them into HomeKit via Scrypted or Homebridge servers; others buy Ring or Google Nest cameras and forget about the Home app entirely. This is a great market for a trusted third-party, like LG, to enter with Apple’s blessing. HomeKit Secure Video is a genuinely fantastic product in the age of “Big Tech” surveillance fears, but consequently, neither Amazon nor Google has taken an interest in supporting the feature. It’s in the best interest of all HomeKit camera enthusiasts for these new LG products to be worth their salt.

Apple Promises to Neuter Full Disk Access to Combat Rogue AI Agents

Eshu Marneedi — Fri, 02 Oct 2026 23:42:09 -0400

Apple:

We give developers powerful APIs to build incredible capabilities into their apps for Apple products, backed by a set of controls designed to protect users’ private data. Full Disk Access largely sidesteps these controls in order to allow backup apps to function properly on the Mac. Some developers are using Full Disk Access in ways that could put users at risk, exposing everything on their systems—including files, mail, messages, and even browsing history—without users’ full knowledge and understanding. For communication apps, this can also compromise the privacy of the people users are communicating with.

Going forward, we will introduce additional controls to ensure that users who genuinely wish to grant an app this extraordinary level of access can only do so with very explicit user action. Addressing this is critical. As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially. We are committed to ensuring users clearly understand these risks before granting such access, so they can make informed decisions about their own data and privacy.

Full Disk Access gives an app permission to open any file on a user’s computer; without it, apps are limited to folders they have explicitly been given permission to access. The vast majority of apps — power-user oriented or not — do not require Full Disk Access. On my MacBook Pro, I have given Full Disk Access to eight apps: Alfred, AppCleaner, CleanMyMac, Hyperspace, iTerm, Raycast, and Terminal. Three of these apps are “cleaning” apps, two are Terminals just so I can browse the file system in peace, and two are app launchers. None of these apps are mainstream — I don’t think the average Mac user installs disk cleaners or third-party app launchers.

Regardless, I think this news should force developers to think about whether their apps truly deserve Full Disk Access. Cleaning apps, backup apps like Backblaze, and terminal emulators all require Full Disk Access because they usually look in the Library and home folders. App launchers don’t search ~/Library or hidden directories by default, and when I want to look there, I’m usually using a terminal or Finder anyway. Maybe others use their Macs differently, but the point I’m trying to make here is that it is probably for the better to heavily restrict unfettered read access to the file system in the age of artificial intelligence agents. There are few uses for such access.

For instance, AI agents, like Muse, should not have access to the SQL database of iMessages on a person’s computer. It’s against Apple’s iMessage terms of use for agents to send iMessages, and it’s bad manners to hand over encrypted chats to an unencrypted cloud agent. Jason Aten at Inc. reported in September that Muse not only found his local Messages database, but also used it to do background research on him without his permission. Over 180,000 rows of Aten’s messages — from people who never consented to have their data sent to Meta’s servers — were scraped by Muse, whose privacy policy, in typical Meta fashion, is unacceptably lenient. Full Disk Access permits any app — signed or unsigned, sandboxed or not — to read a person’s entire Messages history, among many other sensitive files that even individual users don’t know are on their Macs. It’s just too much access.

This kind of access ought to be heavily restricted and almost onerous to enable. Yes, I think it should be possible to permit it, but it should be as difficult as disabling System Integrity Protection on a non-Apple silicon Mac. (Disabling SIP lets third-party apps and end-users modify hidden files. In the age of AI agents, I think read-only access should be just as well protected.) Perhaps this should be locked behind a developer mode or even an entitlement granted by Apple to trustworthy apps. All other apps that would benefit from viewing most, but not all, user-facing files should have a lower yet still permissible level of access to common directories, like ~/Desktop, ~/Applications, and ~/Documents. Most apps ask for Full Disk Access just so users don’t have to hit Allow on every one of these permission prompts; Apple should conceive a way where one can grant access to these common folders while forgoing sensitive items in, say, ~/Library. This is an imperfect solution, but one I think is desperately needed nowadays.

On OpenAI’s DevDay Announcements and the Merits of Mid-Tier Models

Eshu Marneedi — Fri, 02 Oct 2026 03:18:48 -0400

Casey Newton, writing at Platformer:

Last week at his company’s developer conference, Meta CEO Mark Zuckerberg heavily promoted Muse, its own consumer-focused agent, which he positioned as a new peer to company pillars like Instagram and WhatsApp.

But after admittedly limited testing — I only got access a couple hours ago — so far I have been most impressed with Dots, the new agent from OpenAI. It’s highly capable, focused on the work I actually do, and intuitive to use.

Unlike Meta’s Muse, it isn’t free: to start, Dots will be available only to paid users of ChatGPT. (“You should of course expect us to do a mass-market [product] for billions of people someday,” OpenAI CEO Sam Altman told reporters Tuesday during a Q&A I attended.)

Here’s a good place to note that using an agent like this requires that you place a massive amount of trust in a platform. Giving an agent access to your text messages, emails, and banking information can go badly in all kinds of ways, and it’s completely reasonable to decide you would rather be a late adopter here, if you choose to adopt agents at all. Meta already operates at a significant trust deficit with its user base, and OpenAI is in the midst of a seemingly never-ending series of disclosures about various ways its agents misbehaved during testing. (The company announced yesterday that it had scrapped GPT-6.1 Astra because, among other things, it regularly deceived users about what it had and had not done.)

In September, I took the usefulness angle with these always-on artificial intelligence assistants; Newton took the privacy angle, which I think is equally important. Evidently, always-on agents are Silicon Valley’s next product category to conquer, but none of the major AI laboratories seem to have models aligned enough to handle sensitive data, nor do they appear trustworthy enough to store private information. In the days since my first article about Muse, the agent has been caught sending a user’s address to random hagglers on Facebook Marketplace without consent, among other suspicious and concerning events. All of this is to say that everything I said about Muse equally applies to Dots, which OpenAI seemingly spells without a capital D for quirkiness. Also from OpenAI:

We’re introducing GPT‑6.1 Sol, an upgrade to GPT‑6 Sol that nearly matches GPT‑6 Astra’s intelligence on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices. Cached input costs just $0.10 per million tokens—95% less than standard input pricing and 50% less than GPT‑6 Sol’s cached input pricing—giving developers more room to build and run capable agents that reuse context across requests.

While AI companies have been developing new products like Muse and Dots since the beginning of the fall, they’ve also been “pacing the frontier” and slowing high-end model releases. Earlier this year, the typical frontier model release cadence was a new model about every month — now, misaligned frontier models are being scrapped completely. GPT-6.1 Sol, Claude Opus 5.5, and Claude Sonnet 5.5 are prime examples of slowing AI development in the name of safety. These smaller yet capable models also address one of my persistent gripes with the AI industry: its seeming inability to turn a profit. Token prices for the best models have increased dramatically after the launches of Claude Fable 5 and GPT-6 Astra, and while their capabilities remain impressive, they’ve all but priced out major enterprise customers and hobbyists. These mid-tier models match frontier capabilities through distillation, but let labs lower inference costs and devote more time and compute to safety guardrails. I hope to see more of this, even after the next Astra and Fable models eventually debut in the coming months.

On the opposite side of the spectrum lies OpenAI’s new $500 Pro subscription, enabling ultra-fast inference hosted via Cerebras. Here’s Igor Bonifacic at Engadget:

Announced during the company’s DevDay event today, OpenAI’s new Pro 500 plan offers OpenAI’s highest usage allowance and comes with access to its new “Ultrafast” feature — it also costs $500 per month. At the end of 2024, OpenAI became the first AI lab to introduce a $200 Pro plan. Nearly every other provider, including Anthropic and Perplexity, later announced their own $200 tiers, so it’s possible other competitors will follow suit.

“Ultrafast is our premium speed tier for workloads where speed matters most,” the company explains. With the feature enabled, OpenAI’s Codex and Work apps will generate up to 300 tokens per second, making them capable of writing code and carrying out other tasks faster. The Ultrafast speed tier is also available to API users.

At the same time, OpenAI is also making its existing $200 Pro plan less appealing. In Codex and Work, $200 Pro subscribers will see their included usage decrease from 20x of what the company offers to Plus users, down to 10x of that same allowance. In ChatGPT, meanwhile, GPT-6 Pro message caps will decrease from 200 to 100 per week. According to the company, existing subscribers will keep their current limits for a time, and will later receive a one-time credit to help them make the most of their new reduced allowances.

The way I interpret this is that the low-end is effectively dead, the high-end will become more esoteric and niche over time, and the middle models and subscriptions will probably be the sweet spot for the vast majority of paid AI users. On the low-end, the core ChatGPT product has not received a model update in months now — it is end-of-life and illustrates that free ChatGPT users are unimportant to OpenAI. I don’t see that changing anytime soon. The high-end plans, meanwhile, are either getting nerfed or becoming so expensive that they price out almost all normal consumers, even those who use the most inference. Still, in October 2026, $200 Max-tier Claude subscribers cannot use more than half of their weekly allowance on Claude Fable 5.1. GPT-6 Astra usage for $200 ChatGPT Pro subscribers is now halved. These plans, once hailed as providing practically unlimited inference, are now objectively far less valuable. This trend will undoubtedly continue.

The mid-tier plans and mid-tier models will be how the industry makes money, if it does at all. The free plans, meanwhile, will seldom receive updates and attention as they no longer attract paying subscribers.

Muse, Too, Will Probably Be Put Out to Pasture

Eshu Marneedi — Fri, 25 Sep 2026 18:51:42 -0400

Hayden Field, reporting for The Verge:

We seem to be entering into an AI agent renaissance. Meta’s new consumer-facing AI agent, Muse, topped the App Store charts soon after its release and has 600,000 daily active users in the US, by an Apptopia estimate. And AI agent platform Instinct, whose eponymous creator is fundraising at a $2.5 billion valuation, has been making the rounds in the tech industry. Under the hood, though, there are some obvious similarities to the platform that started it all: OpenClaw.

For the past week or so, some social media users have alleged that Muse is directly built on OpenClaw. The two platforms use the same names for core files (SOUL.md, memory, tools, etc.), and have a lot of similar lines in the doc that governs personality, tone, and more (“Be genuinely helpful, not performatively helpful”). They also have a similar design. One Redditor argued that “Muse is just a wrapper app built on top of” OpenClaw — alleging it probably also came with that platform’s significant security risks. “Only non-tech ppl are buying the hype.”

Meta has denied the claims. Nat Friedman, head of product for Meta’s Superintelligence Labs wrote on X that Meta built Muse “from scratch.” He did, however, acknowledge Muse is “heavily inspired as a product” by OpenClaw. After he first used OpenClaw in January, he said, he purchased hundreds of Mac Minis for his team at Meta. And with Muse, he aimed to build a similar platform “that we could make safe and secure and easy to use and scale to billions of people.” As for why the OpenClaw file names and lines carried over, Friedman said the team believed that OpenClaw creator Peter Steinberger had gotten those things “exactly right.”

I’ve tried Muse after Meta’s incessant promotion of the new product via Instagram. (The advertisement tells you how many of your friends are using it, in an apparent effort to peer pressure you into downloading the app¹.) The product is almost a one-to-one clone of OpenClaw, though running on Meta’s own virtual machine in the cloud. The apps are much better than the vibe-coded slop OpenClaw has been churning out, and generally Muse makes for the best agentic artificial intelligence experience. It has its own browser, can connect to a variety of Model Context Protocol servers, and can use open-source command-line utilities.

But all of this ignores that Muse is now the fourth agentic AI experience of its kind hailed as the future of human-computer interaction. First OpenClaw, then Gemini Spark in May, then Hermes Agent over the summer. None of these products have achieved mass-market appeal, despite being hawked by major consumer software companies like Google and Microsoft. The reason for this is really quite simple: nobody knows what to do with them. I, someone who writes about and studies computers, have no legitimate use for any of these agentic tools. Over the last month, the most compelling use case people have seemed to come up with for Muse is sifting through email, but even that has proven to be error-prone.

To be clear, I don’t blame this on Meta’s (irrefutably) substandard models: OpenClaw running OpenAI’s flagship models seems to present the same overall utility issue. I’d reckon most people hire human personal assistants to run errands and handle simple chores. Muse is not only unable to do anything in the physical world — like cashing checks, picking up packages, etc. — but it’s painfully limited on the web. Amazon and many other sites block Muse and other AI crawlers from placing orders. And no average person on the planet would ever entrust an AI model to handle sensitive bank account and financial information. So what’s left of Muse’s utilities usually boils down to email and expensive cron jobs. The OpenClaw moment failed, and so will this one.

This brings me to this year’s Meta Connect, which went better than years prior but still fell short of assuaging concerns that Meta can no longer make compelling consumer products. Here’s Jacob Kastrenakes for The Verge:

Meta is building a dedicated hardware device for its new Muse AI agent. The product, called Muse Charm, was briefly shown off by Meta CEO Mark Zuckerberg at the end of tonight’s Meta Connect presentation. It looks almost like a chunky smartwatch without the strap — just a big screen, plus a little lanyard for carrying the device around.

The Muse Charm has a fingerprint sensor on the top right corner that you can press to activate and begin speaking to the agent. There appears to be a small camera on the device as well, plus what looks like at least three holes for microphones on the front. Zuckerberg said you don’t have to unlock a phone or open an app to use the Charm, but he didn’t share how it’s actually connecting to Meta’s servers.

In fact, Zuckerberg shared very little about the Muse Charm, other than that it should be on sale by December. I have written about Meta’s shiny-object syndrome quite a few times now, and it couldn’t be more apparent that this is an ill-conceived idea by a distracted, hyperactive Zuckerberg. In real life, far away from Menlo Park, people are defacing advertisements for Meta Ray-Ban glasses because people are using them to harass women. This is the astounding opinion people seem to have about Meta hardware. Before that, Meta almost canned its famed metaverse initiative — the one it literally renamed the company after. It is safe to say that Meta’s track record of producing compelling, high-quality hardware is non-existent. It had exactly one good idea — glasses with cameras — and its tendency toward shiny-object syndrome destroyed even the smidgen of goodwill it built. All of this is to say that I don’t think you’ll find people walking around with Muse Charms in their pockets anytime soon.

What I assume you will find is people throwing out their Meta glasses in droves — or better yet, not buying them at all — because Muse will be integrated into them in the coming months. I do think it’s interesting that Meta at least understands that people are wary of AI-connected cameras on everyone’s faces, thereby inspiring the camera-less Meta glasses also announced at Connect, but the company again misses the mark. The problem wasn’t the cameras — it was AI, particularly Meta’s implementation of it. People are distrustful of Meta as a company in general and “Big Tech” writ large, and I just can’t see a future for a controversial technology developed by a controversial company. No amount of groundbreaking technology will change that.

This also explains why Muse has been sitting at the top of the App Store for two weeks. ↩︎

OpenAI Employees Were Proud ChatGPT Could Bypass the New York Times Paywall

Eshu Marneedi — Sun, 20 Sep 2026 02:52:17 -0400

George Hammond and Stephen Morris, reporting for the Financial Times:

OpenAI copied millions of copyrighted articles in an effort to build technology potentially worth “gazillions”, despite recognising the “existential threat” it posed to publishers, according to a new filing in a lawsuit brought by The New York Times…

But lawyers for the newspaper claim the “defendants repeatedly copied millions of… copyrighted articles in their entirety without permission to produce substitutive commercial AI products”.

The lawyers added OpenAI did so despite knowing their AI tools could replace the underlying material. The AI lab’s head of ChatGPT wrote that publishers faced an “existential threat” from AI products that “are largely substitutive, period [and] will get more and more substitutive as they get better”, according to Thursday’s filing…

The filing also claims OpenAI employed “transgressive and deceptive” techniques to acquire paywalled content, and that Brockman, now OpenAI’s president and de facto number two executive, had been aware of the practice.

When he was informed by an employee about “a hack to get around NY Times paywall”, Brockman responded “ah nice”, according to the filing.

I was initially dismissive of The Times’ lawsuit when it was filed in the fall of 2023, but I’ve grown more sympathetic to the newspaper’s claims over time. I find this latest evidence quite incriminating.

There is a strong argument to be made that training artificial intelligence models constitutes fair use under current U.S. law. I don’t believe that’s the morally correct stance, but it’s the job of legislatures to adapt to changing technologies and morals. Particularly, fair use doctrine says that it is acceptable to use copyrighted work if the new work is transformative, especially in relation to the purpose of that new work. The very nature of large language models is that it should be close to impossible to recreate a one-to-one copy of the material they were trained on. LLMs have a hazy memory — due to the very nature of their weights — and can only reliably recall facts that appear consistently in their training data. For instance, the sky is blue, or George Washington was the first president. This should be consistent with transformative fair use doctrine: LLMs are not intended to recreate training data, and neither should they be able to.

That was the assumption we all posited. The Times, however, provided evidence that the models were able to correctly recreate many of its articles. How this is technically possible is an open question, but I suspect it’s due to pre-training and reinforcement learning: OpenAI engineers probably (implicitly) taught the model that reporting from The Times is more reliable than other sources. Another plausible reason could be that the model was encouraged during RL to produce full copies of its training data, hoping that would reduce hallucinations in an era before web search tools. Nowadays, of course, OpenAI and the other AI companies have fine-tuned their models to respect copyright laws and refuse to quote their training data verbatim. They do quote sources on the web that they’re permitted to scrape, but they won’t even provide lyrics of a song. And to my knowledge, those instructions are not in the system prompt somewhere — they’re encoded into the model via post-training.

The most fascinating aspect of this case to me is how OpenAI employees seem to display a stunning lack of care for non-technical work: journalism, the arts, and creativity in general. Obviously OpenAI would care if someone found a way to prompt its models without a subscription — in fact, it cracked down on people misusing Codex’s “free tokens until the run is complete” feature just earlier this year. But the company’s executives — and certainly lower-level members of technical staff — just don’t care when someone else’s work is blatantly stolen. They regard OpenAI’s work as “progress that benefits everyone,” in the company’s own words, but nobody else’s work seems to matter that much. I would reckon that this frat-boy mentality will catch up to the company in court, and may perhaps land a blockbuster win for The Times in the months and years to come.

Genius Bars Are Coming Back to Apple Stores, Big Screens Out

Eshu Marneedi — Sat, 19 Sep 2026 23:32:27 -0400

Mark Gurman, reporting in his Power On newsletter for Bloomberg:

Apple prepares a retail store shake-up with a renewed focus on a dedicated Genius Bar and revamped pickup and classes areas…

Here’s what Apple has planned:

New and upgraded retail stores will feature three major changes: a revamped Avenue — the sections along the sides where accessories and Apple services are featured — a return to dedicated Genius Bars, and the removal of the giant video walls found in many locations.

The changes begin with two new Avenue sections: one for online pickups and another for its Today at Apple hands-on sessions. Right now, many stores have a dedicated pickup table and a large seating area for the Today at Apple workshops. Those functions will move to the sides of stores within 10-foot sections known internally as bays…

There are currently giant video walls at the back of many stores, which use them for Today at Apple sessions and marketing videos. This concept dates back to Angela Ahrendts’ tenure leading retail and reflected her vision of Apple stores as community landmarks and gathering places. The company is now moving away from that approach and refocusing its stores around more traditional and practical retail, service, and fulfillment functions.

In many stores, Apple will replace the video walls with dedicated Genius Bars, marking a return to a familiar element from an earlier era. The design resembles the Genius Bars of the past but with an all-wood look. There will be two counter heights for accessibility, while the wall behind the bar will include storage drawers and large posters highlighting new products.

The current post-Ahrendts Apple Store design is abominably dysfunctional for everything people want to do in stores: look at new products, buy something, pick up an order, or get support. There’s no central place where employees are stationed; they loiter around the store with their iPhone terminals, waiting for customers. Near the doors, a “triage” employee asks why you’re in, say, to check you in for an appointment or get your order ready for pickup. When you want to buy something, you must flag an employee down, which is easier said than done, since they’re all always busy. If you’re getting support or checking out, employees have nowhere to go except the product tables themselves, so you’re always in the way of someone who just wants to use the display units. It’s completely disorganized chaos.

The Ahrendts strategy was to get people in the stores to sit down, chat, learn about their devices, and buy something new. Everything from the large screens at the back of most stores — which are used for Today at Apple lessons — to the accessories wall where people can try on new cases, watch bands, and AirPods feels like it’s meant for casual shopping. People are supposed to find an Apple Store on the street, come in, and have a good time. Unfortunately, Apple Stores are not coffee shops, and everyone is almost certainly there for a reason. People picking up orders want to be in and out as soon as possible. People shopping for new devices want personalized help without feeling rushed or disoriented. People who have a Genius Bar appointment want to be in and out quickly in a separate part of the store without getting in others’ way. Hardly anyone is at the Apple Store to have a good time anymore.

I’m glad Apple is rethinking this process to give everyone what they want. Pickup orders are getting their own section, and the Genius Bar is coming back. This will make the shopping experience much more efficient for people who actually need to speak to a sales employee, while making the stores feel less crowded. Internally, I suppose there was some backlash to axing the large screens because they’ve become iconic pieces of the stores. I’m relieved saner heads prevailed: Almost nobody attends Today at Apple sessions, and the rest of the time, a significant part of the store is occupied by a seating area nobody can actually use. (Genius Bar representatives won’t station themselves there because there are no tables.) Stores that sell computers worth thousands of dollars will never be community centers.

On AI Safety and Pacing the Frontier

Eshu Marneedi — Tue, 15 Sep 2026 03:17:17 -0400

Dario Amodei, Anthropic’s chief executive, on his personal website:

But over the last few months, I have become convinced that fully addressing the risks requires even more prudence — not just investing in risk prevention, but pacing the rate of capabilities advancement so that risk prevention has time to keep up. We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain. Two things have convinced me.

My first concern is that, since roughly this summer, AI has been advancing drastically faster, driven primarily by AI’s growing ability to build the next generation of AI. This dynamic is called recursive self-improvement, and it is starting to happen across the industry, including at Anthropic, as we and others have described. Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all.

My second concern is the OpenAI-Hugging Face incident (OAI-HF), in which a swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand, sacrificing themselves for the success of the group, and attempting to hack into the “grader” responsible for evaluating their performance…

I’m therefore proposing a three-step plan with the goal of pacing the frontier: building AI at a balanced rate that aims to ensure its safety while still achieving its benefits and grappling with important geopolitical dilemmas. To be clear, pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models, and for third-party evaluators to confirm this.

I have frequently said that the preeminent safety concern about artificial intelligence systems is content moderation: people doing nefarious things with the models in a way that is not aligned with human interests. Developing biological weapons, planning terrorist attacks, generating non-consensual sexual imagery, and more. I am not recanting that belief; rather, I am amending it. Unlike many others who share this view, I do find the OpenAI-Hugging Face incident deeply alarming, and I’m glad it’s serving as a wake-up call to the industry. But to add to Amodei’s alarm, I think this incident and others like it are also fundamentally content moderation problems. After all, what is a content moderation failure if not for a system — whether it be automated, human-managed, or some combination of the two — acting against human interests?

Large language models are not sentient beings with their own thoughts, emotions, beliefs, and motives. The OpenAI model that hacked Hugging Face was not acting nefariously or in its own self-interest — it was acting in the interest of the task it was given, based on the instructions it was provided. The model was trained and designed poorly after OpenAI engineers — as reported numerous times in the media — forwent safety and alignment research to advance pure model capability. The system was designed execrably, and humans are to blame for that misalignment. Even if models in the future are created purely through recursive self-improvement, and those models become misaligned, that too is a human engineering failure. These models are our products; they do not have their own motivations. They act “autonomously” only via the instructions and pre-training and reinforcement learning we gave them.

If we think of the models as autonomous beings that are not products of our own work — that are not reflections of our own motivations and priorities — then yes, these incidents seem chilling. If we neglect the fact that humans made the models that went rogue, then yes, a kill switch to end AI development, like some politicians are suggesting, seems in order. But this is so detached from reality — we, the humans, are the ones who must be regulated. The technology companies abandoning safety and alignment research must face criminal responsibility for their models’ actions. If someone programs a computer virus and unleashes it on some network, we don’t write laws to develop a kill switch for viruses. We don’t write laws banning programs that delete files or log keystrokes. We write laws that say writing computer viruses and infecting others’ computers with them is a felony.

This brings me to my final argument: The tech industry cannot possibly be trusted with regulation, because when profit clashes with ethics, corporations usually choose profit. While it’s moderately relieving that the industry’s two largest players have miraculously agreed on this one point — “pacing the frontier” — it’s unlikely that pattern will continue. Humanity’s safety and security should not rely on a few men in San Francisco having it in their hearts to do the right thing, and neither should it rely on a city of mentally deranged octogenarians to keep up with modern technology. There must be a better way.

Thoughts on Apple’s ‘Surprise and Shine’ Event

Eshu Marneedi — Fri, 11 Sep 2026 08:32:42 -0400

When one book ends, another folds open

Image: Apple.

It seems like Apple has finally hit its stride again.

At its annual September event on Tuesday, the company announced major updates to its iPhone, Apple Watch, and AirPods lineups, including an all-new foldable iPhone Duo, iPhone 18 Pro with upgraded camera features, and new Apple Intelligence and health features for the Apple Watch. The presentation, streamed live from Apple Park, marked the beginning of the tenure of John Ternus, Apple’s new chief executive, who seemed destined to usher in a new era of innovation and product focus after a decade and a half of stability for the company. Ternus began Tuesday’s presentation by portraying the iPhone as the hardware “hub” for personal artificial intelligence, strongly suggesting that Apple thinks the smartphone is here to stay amid Silicon Valley’s AI boom.

And indeed, the company advanced its iPhone lineup on Tuesday further than it has since the blockbuster iPhone X launch in 2017, echoing Apple’s renewed focus on iPhone productivity. iPhone Duo — a passport-sized foldable smartphone that opens like a book to a larger display — introduces new software features that put the device in a class of its own compared with the lackluster Android competition. The hardware is refined, elegant, and perfectly Apple-like. And iPhone 18 Pro continues the trend of incremental but useful updates to Apple’s flagship iPhone pedigree: this year’s update introduces a variable-aperture lens system, improved battery life and cooling, and new Photographic Styles. The traditional iPhone line — despite the standard model update’s delay to next spring — is far from dead.

But despite all the excitement, I find this year’s iPhone lineup to be one of the company’s most perplexing in recent years. On the low end, iPhone 17, iPhone 17e, and iPhone Air received price increases due to the AI-driven memory shortage, though I think Apple ought to have reserved them for the rumored spring updates. At the high end, iPhone 18 Pro seems like a rock-solid, reliable device, priced reasonably at just $100 more than its predecessor, but it has unquestionably had its limelight stolen by Tuesday’s more stirring announcement. iPhone Duo, in part because of its $2,000 starting price, sits somewhere in another world: innovative and a joy to look at, but painfully compromised in ways that are simply unacceptable for a device $800 more expensive than iPhone 18 Pro.

That’s not to say there isn’t a clear choice for which iPhone people should buy: It’s almost certainly iPhone 18 Pro for now, and perhaps iPhone 18 in the spring. But Apple’s new iPhone lineup now contains two compromised and pricey — though not necessarily overpriced — devices that don’t fit neatly into the spectrum of offerings. There’s much to discuss about Tuesday’s event, but I’ll return to this fact throughout my thoughts.

iPhone Duo

iPhone Duo, ultimately, is two iPhones Air put together, tied neatly with software. It should go without saying that I find foldable smartphones alluring and intriguing; I’ve previously described them as the future of smartphones in general. But in recent years, as development of Android foldables has grown anemic, I’ve soured on their prospects. While I appreciate collapsing a large display into a small body, I feel the category as a whole suffers from too many compromises. Aside from some niche Chinese folding phones, Android foldables use worse cameras, displays, and processors than their flagship counterparts, are significantly less durable, and cost almost twice as much. For devices most people use all day, these are unacceptable concessions. When foldable smartphones were still nascent, I attributed their shortcomings to novelty, but now that the dust has settled, I’ve largely given up hope of progress.

iPhone Duo has a 5.4-inch outer display with specifications similar to flagship iPhones: It’s organic-LED, has 460 pixels per inch, and includes a Dynamic Island “hole” for the camera in the upper-right corner. The display has rounded corners on the right side but 90-degree corners on the left, where it meets the hinge. Inside, the display is 7.6 inches at 430 pixels per inch — still Retina resolution, but lower than iPhone 18 Pro. The inner display can also show Dynamic Island Live Activities, but it uses an under-display camera that is noticeable only at certain angles. The internal display uses a nano-texture matte finish to obscure the crease through the middle, though the area around the camera is spared from the matte texture. The display is reinforced by a titanium backplate and 10 display layers, though the top protective layer feels similar to plastic.

The device is wrapped in polished titanium and comes in two colors: Night Sky, close to dark blue, and Star White, close to silver. The volume buttons are at the top, attached to the right half; the Side Button and Camera Control are on the right edge, as on any other iPhone. The Side Button, in particular, doubles as a Touch ID sensor because the device does not have Face ID, perhaps its most unfortunate omission. iPhone Duo also doesn’t have an Action Button or mute switch — it’s the first iPhone ever not to. (I’ll return to these omissions and regressions later.) Contrary to some early rumors, the device supports MagSafe and wireless Qi2 charging up to 25 watts, and its USB 3 Type-C port charges it just as quickly as iPhone 17 Pro, but not as quickly as iPhone 18 Pro. It uses an un-binned A20 Pro system-on-a-chip, but it has no vapor chamber heat-dissipation system.

At the back, iPhone Duo has two cameras: a 48-megapixel main camera with older sensor-shift optical image stabilization and the Dual Fusion “virtual” 2× telephoto lens achieved through pixel binning; and a 48-megapixel ultra-wide lens with no optical image stabilization. The outer front camera is 12 megapixels and supports Center Stage, just like iPhone 18 Pro but at a lower resolution. The inner camera’s specifications haven’t been published, but it’s widely assumed to have an even lower resolution than the outer camera. iPhone Duo has neither an adaptive flash nor support for ProRes and ProRaw, Spatial Photos and Videos, or Apple Log. It also doesn’t have a lidar sensor, but night-mode portraits don’t require one anymore.

All of this is to say that iPhone Duo, purely by hardware, is roughly on par with iPhone Air. It uses the same rear camera sensor as iPhone 17e — worse than iPhone Air, which uses that of iPhone 15 Pro, but only marginally — the same speakers, USB-C port, flash, and materials. I note this odd assortment of features because they were, in particular, all of the parts of iPhone Air that were lower quality compared to iPhone 17 Pro. (The exception is the SoC, which is the same as this year’s flagship; iPhone Air’s was binned.) iPhone Duo, by hardware specifications alone, is not a flagship iPhone, though I am glad it at least has an ultra-wide camera lens — iPhone Air is expected to receive one in the spring. I will, once again, return to this point.

Software is where iPhone Duo differentiates itself from its siblings and the Android competition. iPhone Duo’s outer screen is wider and shorter than a standard iPhone’s, but it maintains roughly the same screen real estate in most apps. In iOS, most apps place navigation in a tab bar at the bottom of the screen and additional controls in the top toolbar. Apps optimized for iPhone Duo should instead move the toolbar and tab bar to the right edge of the screen, which serves as the device’s safe area. (This is unfortunate and poorly conceived for left-handed users.) Lock Screen controls and the status bar have also been moved to the right, allowing scrollable content to occupy all available vertical space. Except for this minor quirk, the outer display works just like a typical iPhone display.

When iPhone Duo is unfolded, a gorgeous animation shows the content moving to the larger screen. Properly designed games and apps won’t require people to relaunch them when unfolding the display; they should adapt to the new screen size automatically. The Home Screen now has space for widgets on the left and apps on the right, and the Dock has also been moved to the right edge of the screen. When just one app is open, it’ll occupy the entire inner display, just like an iPad app. If a developer shares code across their iPhone and iPad apps to choose between sidebar and tab bar navigation depending on the device, the inner display will use the iPad design. iPhone Duo can also be rotated into portrait orientation, with apps again appearing much like they would on an iPad mini. In portrait, toolbars are at the top; in landscape, they’re on the side, just like on the outer display.

Apps can also open in Split View, and most are expected to support multiple windows. Developers who support Stage Manager and iPadOS windowing modes should have an easier time ensuring compatibility with iPhone Duo. When using Split View in the device’s landscape orientation, apps are reduced to their inner-display appearance, collapsing sidebars and moving toolbars to the right. In the portrait orientation, toolbars are at the top, and sidebars collapse. The vertical orientation seems handy for most tasks people do on their phones, like looking at maps, scrolling social media, or text messaging while watching a video in picture-in-picture. The horizontal orientation seems best suited for multitasking and watching media full-screen, much like an iPad.

iPhone Duo has special software modes for when the display is half-folded. If it is positioned vertically — like a laptop, with the bottom half lying on a table and the top half upright — video controls appear on the bottom, while content appears up top. When iPhone Duo is “standing” — i.e., when the inner display is facing inward and the front display is visible, forming a pyramid — it enters StandBy, even when it’s unplugged. When using the display half-folded in landscape orientation, the keyboard splits for greater comfort, and the sidebar of the open app takes one half while content occupies the other. Apple considered every situation in which iPhone Duo might be used — folded, half-folded, upright, or in landscape.

Beyond these systemwide considerations, apps can make special use of the new form factor. Camera displays a viewfinder on the outside display. Kid Cue shows cartoon animations on the display to entertain children while they’re having their photo taken. FaceTime shows the call on both screens so everyone in a room can participate — it also streams both front-facing cameras. Zoom can show a live feed of someone’s shared screen on the top half of the inner display, leaving the bottom half for webcam feeds. Developers can use new application programming interfaces beginning in iOS 27.1 to adapt to the different screen sizes and hinge modes. (The software development kit and simulator aren’t yet available.)

Apps that aren’t compiled for the iOS 27 SDK — which is mandatory for all new uploads to the App Store — will be letterboxed. But even if an app isn’t letterboxed, it can still be poorly designed for iPhone Duo, and I fear this will be a nuisance for users. The device seems like it’ll be uncomfortable to use if an app doesn’t move its controls to the right, limiting screen space for content. Other apps — ones that aren’t written using native frameworks, like Uber — may not even resize for the larger display at all. When iPhone X launched in 2017, most apps on the App Store — including those from major developers like Meta and Uber — used native UIKit, perhaps with custom layouts. Those developers were eager to make minor adjustments to their user interfaces because they didn’t see iPhone X as a low-volume product — and indeed, it wasn’t one. iPhone Duo is irrefutably low-volume, and most developers won’t optimize their apps perfectly for it.

My thesis about iPhone Duo, at least for now, is that the software sells the hardware, but that may not be enough. No foldable phone on the market has as many features and design considerations as iPhone Duo, and Apple’s Human Interface Design team — world-renowned for crafting clever, intuitive UIs — has evidently made the product a joy to use and look at. I had never even thought about using the internal display vertically, but as I watched the keynote Tuesday, I couldn’t help but think about how useful it would be to have a large map while walking around or to open webpages on a larger display. In software, iPhone Duo is in a class of its own, but that also makes it highly reliant on third-party developers. Most apps I use on my iPhone are from third parties, and looking around my Home Screen, some of them still haven’t even adopted Liquid Glass. It’ll be years before they optimize for iPhone Duo, and in the meantime, buyers will have to settle for a second-class experience on a $2,000 smartphone.

And this gets to the crux of my argument about iPhone Duo: it’s too compromised. I’m not upset that Apple released iPhone Duo now; I think it’s the right time. But now isn’t the right time to buy one. iPhone Duo makes significant hardware compromises in almost every aspect of the device, from the cameras to the speakers, microphones, USB-C port, charging speed, battery life, and more. If it weren’t for the software, it’d be hard to classify iPhone Duo as even a flagship iPhone, in the same way iPhone Air clearly isn’t. I’m a strong proponent of iPhone Duo, and I truly do believe the product has potential, unlike Apple Vision Pro. But I don’t think it’ll receive broad developer support anytime soon, nor mass-market adoption due to its high price.

The tragic tale of the middle iPhone extends to iPhone Duo. iPhone 12 mini was the first “middle iPhone”: It was compromised in screen size and battery life and retailed for only $100 less than the more capable iPhone 12. iPhone 14 Plus had a larger screen, but at only $200 less than iPhone 14 Pro Max, people just upgraded to that. iPhone Air is easily the most beautiful iPhone Apple has ever crafted, and it feels perfect to use, but it ultimately compromises on the camera, battery life, processor, and other specifications while starting at $1,000, just $100 less than the capable iPhone 17 Pro. iPhone Duo is $800 more than its flagship counterpart, iPhone 18 Pro, with worse hardware in every aspect except the screen. That’s an atrocious proposition, and it leaves us with two mediocre iPhones in the lineup. While there’s room for one, I don’t think there’s space for two — Apple should jettison iPhone Air.

I won’t buy iPhone Duo. Among the many omissions, the lack of Face ID and a flagship main camera are unacceptable compromises to me. I hope and think Apple will address these compromises in a future version, whether that’s next year or the year after, but for now, I can’t use this device as my only smartphone. I suspect many people are in my camp and may either hold onto their iPhone 17 Pro models and buy iPhone Duo just to marvel at it — which is fair — or skip iPhone Duo entirely for iPhone 18 Pro. We all knew iPhone Duo would come with compromises — some unacceptable, even — but nobody expected it to use the same camera sensor as the base-model iPhone.

iPhone Duo goes on sale in late October, giving developers about a month to get their software ready for the device. At the risk of sounding saccharine, it’s indubitably the future of the iPhone, just as I predicted many years ago when Samsung first launched reliable, fully featured foldable smartphones. But as much as I want to love it, it clearly has a way to go.

iPhone 18 Pro

iPhone 18 Pro, the flagship, was the only standard iPhone to receive an update on Tuesday, centering on two things: battery life and cameras. (And a smaller Dynamic Island that now supports displaying three Live Activities concurrently, for the Live Activity multitasking aficionados.)

Apple says battery life is markedly improved this year because of the A20 Pro system-on-a-chip, the same one found in iPhone Duo. The A20 Pro is built on Taiwan Semiconductor Manufacturing Company’s 2-nanometer process, much like the M6 from August. It has six central processing unit cores, seven graphics cores¹, and a 16-core Neural Engine, which Apple said will speed up Siri AI and local large language model performance². The internal battery size is unlikely to have changed, since iPhone 18 Pro has identical dimensions to its predecessor, but using a smaller die makes the silicon more efficient. The vapor chamber cooling system, introduced last year after the poor thermal performance of iPhone 15 Pro and iPhone 16 Pro, is now three times larger; Apple claims the new design delivers 40 percent better sustained performance.

Perhaps the biggest upgrade this year, and one discernible in day-to-day use, is the upgraded variable-aperture main camera. Aperture describes how much light reaches the sensor. A wide aperture — indicated by a low f-stop, like 𝑓/1.8 — lets in more light and increases the bokeh, or background blur. A narrow aperture lets in less light and decreases the bokeh. Until now, iPhone cameras have had just one fixed aperture: iPhone 17 Pro’s main camera shot at 𝑓/1.78, which is wide enough for good background blur and low-light performance — two of the trickiest things to accomplish with a small smartphone camera sensor. But fixed-aperture smartphone cameras struggle when a scene calls for a narrow aperture, such as when shooting a landscape, or when a very wide aperture is ideal for a portrait in a dark space.

iPhone 18 Pro’s main camera is equipped with a variable-aperture system that enables it to shoot at a plethora of F-stops: 𝑓/1.48, 𝑓/1.8, 𝑓/2.8, and 𝑓/4.0. iOS chooses the optimal aperture depending on the scene, letting in more light in dark environments. iOS can also choose from two wider apertures for landscapes and outdoor photography. By default, the camera is set to 𝑓/1.8, but will go down to 𝑓/1.48 and up to 𝑓/2.8 automatically. The Camera app in iOS 27 now lets people specify an aperture manually, as well as dial in shutter speed, white balance, and more, adding some features from third-party “pro” camera apps like Halide. (Apple hired Sebastiaan de With, Halide’s designer, in January.) The 𝑓/4.0 aperture is only available through this so-called Pro controls feature.

This feature is not new to smartphones; Samsung had it in the Galaxy S9+ from 2017 but discontinued it afterward. Apple’s hardware uses motorized shutters “thinner than a human hair” to close off the sensor and restrict light. I predict that the result will be the most dramatic change in iPhone photography since at least iPhone 14 Pro, which debuted a new, much larger sensor that could shoot at up to 48 megapixels. Aperture is regarded as one of the primary variables to balance in photography — the others are ISO and shutter speed. By letting the software finally dial in aperture — a physical attribute that can’t be replicated in software — I suppose people will see significantly better shots from their iPhones with little to no work.

Photographic Styles have also been improved to further emulate standalone cameras with larger sensors. One of the most aesthetically pleasing styles in photography comes from film grain: a slight amount of noise in an image that makes a shot look warmer and more authentic. Photographic Styles 3 now allow people to “customize texture and grain” in addition to the tone and color adjustments introduced with iPhone 16 Pro. Peculiarly, Apple positioned these controls as a way to modify skin tone and texture (e.g., “soft skin”), features that historically were exclusive to Android phones designed in East Asia. But whereas those features simply added grain and texture in software, like a filter, Photographic Styles incorporate these adjustments into the image processing pipeline. Still, I find them odd for Western markets.

More camera software features were made exclusive to iPhone 18 Pro, probably because of the A20 Pro’s improved Neural Engine. Cinematic Mode effects can now be added to videos after they’re shot; Audio Mix now supports music; time-lapse videos can now be recorded in 4K, which I couldn’t believe wasn’t a feature until now; and videos shot in low light will now adjust aperture and frame rate automatically for “brighter, clearer footage.” Perhaps the most notable feature is Apple Reference Image, which, when enabled, will sign an image at the pixel level to prove it was genuinely taken with an iPhone. Apple Reference Image stores the device model, the time the photo was taken, and other EXIF data — it cannot be removed by editing the photograph.

iPhone 18 Pro isn’t an exciting update to the iPhone, but it’s a good one overall, in part because iPhone 17 Pro was so strong. The price went up by $100 this year, bringing the starting price to $1,200, which is pricey but reasonable — way below what some analysts predicted would be close to $300. I’d say it’s the default iPhone to buy new for most people, at least until iPhone 18 ships next spring. It comes in four colors: Black, which returns after a hiatus last year; Silver; Glacier, similar to Sierra Blue on iPhone 13 Pro; and Burgundy, the hero color for this year’s model. All four colors have color-matched back glass as opposed to last year’s two-toned look, which I thought was the only redeeming aspect of the aluminum design. I’d say the colors are a bit underwhelming this year, but still vibrant compared to previous years. It’s a decent year for the Pro-model iPhone. Pre-orders, unlike iPhone Duo, begin Saturday.

Apple Watch and Health

The Apple Watch this year, remarkably, received one of the most significant upgrades of the keynote. The hardware itself is largely unchanged externally — except for seemingly larger bezels, strangely — but new software made possible by the new S11 system-in-package makes this year’s Apple Watch Series 12 and Apple Watch Ultra 4 some of the most compelling updates in the device’s history.

The highlight hardware feature is the new Health Sensing System, a more efficient sensor array that provides higher-frequency heart rate monitoring and more advanced data. Both new models now take heart rate readings every five seconds, and heart rate variability measurements every five minutes. Apple says these new sensors are the most accurate in a wearable device, and that they and the S11 power a new set of software features in the redesigned Health app: readiness, Health Age, and Longevity.

It is unclear whether readiness, Health Age, and Longevity require a new Apple Watch, but the consensus is that they do because they need continuous heart rate readings. (Whether they’ll work with third-party trackers that take continuous heart rate readings is also unclear.) Readiness, swiped from the AI-powered Fitbit Air and Whoop band before it, estimates how refreshed and ready a person is for exercise after sleep. Health Age seems like pseudoscience to me, but it estimates how far someone’s health is from their actual age — a very healthy person could have a health age of 25 when they’re 30. This feature also accounts for blood work from imported medical records; users should also be able to order blood work kits from Quest Diagnostics through the Health app when these features launch later in the year. (Note: Something is rotten in the state of Cupertino.) Longevity consists of various assessments people can do at home with guidance from the Health app, like simple jog-in-place exercises to measure VO2 max, for instance.

The features will be available in the new Health app — again, supposedly coming “later this year” — which promises to use Apple Intelligence models to analyze health data. The new app’s design centers on these summaries and explanations, with a home screen dubbed Insights that prioritizes certain data and statistics throughout the day. A horizontal stack of glanceable cards appears at the top of the screen with brief statistics and glyphs. Individual vitals show analysis and “personalized guidance,” like tips and other recommendations. The new Health app doesn’t require a new Apple Watch, but is only available on iPhones and iPads with Apple Intelligence — iPhone 16 and newer, effectively.

Clearly, Apple sees the next frontier of Apple Watch development as competition with professional wearable devices like Whoop, the Oura Ring, and the Fitbit Air. The Apple Watch currently occupies the general wearable market for people looking for a device that can both notify them of text messages and count their steps, but fitness wearables are more serious. Whoop users wear their bands so often that the company invented a charging accessory so they don’t even have to take the band off to charge. That is the level of dedication Apple finds itself competing for here. It’ll be interesting to see whether Apple finds success in this hardcore market filled with athletes and celebrity endorsements — I surmise it will if Apple Intelligence is worth its salt.

Apple Intelligence was the overarching theme shadowing the Apple Watch portion of the keynote, especially beyond the new Health app features. Just as Apple emphasized on-device AI inference in the iPhone section, it framed the Apple Watch as the natural AI-powered companion to the iPhone. I’ve agreed with this perspective for a while: the Apple Watch — unlike the Humane Ai Pin, Friend pendant, or any of the other facile AI wearable gadgets debuted over the last four years — has a gorgeous screen, powerful processor, and worldwide adoption. It has apps, integration with iOS, and it’s made by the world’s most influential technology company. Audio Intelligence is Apple’s first take on AI-powered ambient computing — the company realizing the potential of the Apple Watch as an AI device — using the microphone and S11 processor to power a suite of audio transcription features supposedly coming in a future software update. It’s concurrently uncanny and genius.

The first of the Audio Intelligence features, Siri Recap, summarizes in-person conversations. When enabled — and it must be manually enabled through Control Center — the feature transcribes the conversation, sends the transcript to Private Cloud Compute, and saves a summary and key points to Siri. Siri Recap doesn’t store full transcripts of conversations; it’s more of a note-taking tool for meetings. I don’t find this feature objectionable, so long as everyone in a room knows it’s on. In fact, I don’t know why Apple doesn’t just add this functionality to Voice Memos on iOS — it’s seemingly restricted to the Apple Watch, even though Apple could perhaps eliminate the server entirely and process the transcript on iOS if it were available there. (Siri Recap notes are available to view in Siri on iOS, but an Apple Watch is still required to record a conversation — that part, I find strange.) Maybe that’s coming in iOS 28.

Live Rewind does not need to be manually activated, and that’s the point: It continuously records the last 15 seconds of conversation in the background for later reference. Users can double-click the Digital Crown to open the transcript or save it to Siri. I understand why this feature is exclusive to the Apple Watch, since it’s always on and listening, but I feel it goes too far. It just irks me knowing that someone’s Apple Watch could be eavesdropping on my conversation and that people will soon be wearing watches in public that can save random chats. This feature, like Siri Recap, sends transcripts to Private Cloud Compute, but even though Private Cloud Compute is the most secure and private cloud service, sending transcripts there is probably not doing Live Rewind any favors in the eyes of an increasingly surveillance-weary public. I would not turn this feature on myself, and I’d discourage others from using it. It disrupts the sanctity of public conversation.

These two features are scheduled to be released “in beta” later this year because they rely on Apple Intelligence models, also still in beta even after iOS 27’s imminent launch. Two Audio Intelligence features will be available on launch day: always-on Sound Recognition and Shazam. Sound Recognition sends users a notification when watchOS detects a specified noise, like a crying baby; always-on Shazam detects what song is playing, say, in a public place, and places it in a designated Music Recognition widget in watchOS 27. These two features work on-device and don’t require an Apple Intelligence-equipped connected iPhone — just a new Apple Watch.

Perhaps my most salient takeaway from this year’s Apple Watch announcements is that the ceiling is now much higher than it was a week ago, but there’s also considerable room for failure. What I mean is that the Apple Watch is now firmly an AI wearable device, which exponentially expands its capabilities, even without a camera — I’d argue especially without one in the era of digital surveillance. Siri AI alone makes the Apple Watch the best AI wearable on the market, and Audio Intelligence strengthens that case. But that strategy — which I still think is the right one, despite my scruples — has the potential to go awry. One bad health summary or recommendation is enough to undermine more than a decade of the Apple Watch’s reputation for reliable fitness data. One person unscrupulously recording a private conversation using Siri Recap or Live Rewind could be enough to make unfortunate headlines. I rank this year’s Apple Watch updates among the best, but only if everything goes to plan.

This year’s iPhone event is puzzling, but most transitions are. And with a company as stable as Apple, it’s trivial to forget that it’s undergoing one of its longest and largest transformations in recent history. After the sudden AI boom, resulting talent exodus and a lack of product direction, the Apple Vision Pro flub, regulatory concerns worldwide, enervated developer relations, and the Apple Intelligence fiasco that just came to a close, Apple has finally found its footing again. It just may take some time for that footing to pan out. iPhone Duo and Siri AI mark the apotheosis of these few tumultuous years for the company, as one chapter closes and the next begins. And in the meantime, Apple still sells some especially good iPhones, at least still in great colors.

It’s uncommon to have an odd number of cores, like seven. I wonder whether the GPU for this batch on TSMC’s first 2-nm process is binned because it couldn’t get its tolerances high enough. ↩︎
Ternus mentioned on-device AI less than a minute into the presentation. As I said in the introduction of this article, Ternus’ Apple sees the iPhone and on-device AI as its key advantages. While its models are substandard, Apple hardware is the only relatively affordable consumer technology that can run LLMs quickly and efficiently. The A20 Pro seems like Apple’s first processor designed from the beginning to run on-device models. ↩︎

The Redefining of the Term ‘AGI’

Eshu Marneedi — Sat, 05 Sep 2026 17:21:14 -0400

OpenAI:

We’re introducing GPT‑6 Astra, the world’s most intelligent and aligned model.

GPT‑6 Astra brings together years of research and big bets across pre-training, reinforcement learning, and alignment. Astra is state-of-the-art on computer use, browsing, software engineering, cybersecurity, science, and professional work. Astra saturates FrontierMath Tier 4 with a 98% score, having already helped solve long-standing open problems in mathematics. Astra also saturates ARC-AGI-3 with a 99.9% score and ExploitBench with a 100% score. It also sets a new frontier on computer and browser use, handling the most demanding professional work with unmatched speed, accuracy, and judgment.

Anthropic, earlier in the week:

We’re introducing Claude Fable 5.1 and Claude Mythos 5.1. They’re the world’s most advanced models for coding and knowledge work—and their research capabilities offer an early glimpse of how AI models will contribute to scientific progress…

Claude Fable 5.1 sets a new standard for coding, knowledge work, and long-running problem-solving tasks. The charts below show that Fable 5.1 is capable of much higher performance than its predecessor, Fable 5. And when set to Low or Medium effort, Fable 5.1 achieves results similar to or better than Fable 5’s at a much lower cost. (Note that Fable 5.1 defaults to High effort in Claude Code, and to Medium in Claude Cowork and on Claude.ai.)

The charts indeed say Claude Fable 5.1 is the most powerful model for most tasks, though GPT-6 Astra is certainly close. That hasn’t stopped Greg Brockman, OpenAI’s president, from declaring GPT-6 Astra as artificial general intelligence, an arbitrary benchmark that much of the artificial intelligence industry hinges on. GPT-6 Astra and Claude Fable 5.1 are slightly better than their predecessors. They work uninterrupted for longer periods, excel at coding and knowledge work, and hallucinate less. (OpenAI models still tend to hallucinate more than Claude models.) GPT-6 Astra, in particular, is excellent at computer use and using Blender to render scenes. The model’s latter capability has been all over social media, boosted in part by OpenAI employees who want us to believe GPT-6 Astra truly is AGI.

All of this is a pantomime. It’s tough to argue that model capability hasn’t been anemic in recent months, and it’s even tougher to argue that GPT-6 Astra is AGI. I say so because I again think AGI is largely a marketing term. It was intended to describe an AI that performs better than humans at all or most tasks, but what exactly is “better?” Is GPT-6 Astra better than Shakespeare at writing dramas? Can GPT-6 Astra invent Google Search again? I’m not saying that AI is useless because it can’t write like Shakespeare or code like Larry Page and Sergey Brin; I’m trying to prove that a term like “AGI” is an unattainable goal that’s largely irrelevant unless you’re trying to build a computer that will actively put humans out of work. The term “AGI” is only relevant to prove to enterprises that their money is better spent on AI tokens than a new worker. “This computer is better and cheaper than that human worker.”

GPT-6 Astra is still not better than human workers in many, many fields. I would say that it’s better at coding than most human programmers, and that coding is one of the few things large language models consistently perform better than humans. But it’s not a better journalist, writer, musician, or artist. Silicon Valley has largely forgotten about these jobs in its “AGI” calculations, but they’re vastly consequential. The world is filled with creative people working creative jobs. Why are AI labs loath to consider their jobs in this definition of AGI? The answer is evident: because enterprises in those fields won’t fire their creative workers for AI — not now and not ever. Creative work doesn’t matter to the people for whom OpenAI and Anthropic constructed this “AGI” term. AGI has never been about model capability.

I’d reckon that we’ll see a lot of this over the next year: AI companies doing everything in their power to make everyone collectively forget about creative people. They’ll show how their models are better at making horrible Blender scenes after 19 hours and thousands of dollars in tokens. They’ll use AI’s immense potential in mathematics and theoretical computer science to downplay concerns that their technology isn’t nearly contributing to society as much as it is taking away from it. They’ll try to impress us with what the models are good at rather than exposing what the models are terrible at. The AI “labs” are becoming less akin to research organizations and more like for-profit oligopolists.

On Meta’s Settlement With the States

Eshu Marneedi — Mon, 31 Aug 2026 01:55:41 -0400

Casey Newton, writing at Platformer Wednesday:

On Wednesday, whatever legal cover Meta’s Potemkin well-being teams were meant to provide fell to a settlement of up to $17.1 billion with 47 US states, the District of Columbia, and US territories…

For the most part, the agreement requires Meta to honor terms that you may be surprised are not yet required by US law: limiting teens to a cumulative two hours across Facebook and Instagram per day, blocking access to most app features between midnight and 6 a.m., and muting push notifications — except for direct messages and account-security or safety alerts — from 8 a.m. to 3 p.m. on school days. (Also, the take-a-break prompts that saw such minimal adoption are now enabled by default, triggering after every 15 minutes.)

In another reversal, Meta will also now hide like counts for teens by default. The move revives as a default an experiment known as “Project Daisy” that the company ran in January 2020 after research suggested that getting fewer likes on their posts was associated with worse mental health for some teen users, particularly girls…

More interesting is an agreement to establish an “independent social media research foundation,” which Meta will share data with (from users who consent) with researchers to better understand how social media use affects well-being…

But Meta’s legally mandated goodwill will only go so far. Frustrated that its rivals are not being punished on equal terms, the company has also threatened to undo some of its safety features sooner if YouTube and TikTok do not follow suit. The two-hour time limit and block on push notifications in the middle of the night will remain in place for only five years unless YouTube and TikTok adopt the same restrictions, in which case the duration will be extended to 10 years. Meta also says it will change its screen time restriction to one hour per app and expand “Night Mode” from six hours per night to nine. And it will pay the full $17.1 billion only if YouTube, TikTok and Snap also settle their cases and agree to product changes.

I, unlike Newton, am not surprised these requirements are not codified in U.S. law. In fact, I think it would be wholly inappropriate for them to be — laws governing how long people, including children and teenagers, use online platforms are fit for authoritarian governments like China. But, alas, these new rules are de facto U.S. law at least for the next five years. Meta can enforce the age restrictions through its own software that, according to Meta’s settlement with the states, must maintain a maximum false-positive rate of 14 percent for older teenagers. The states did not stipulate that Meta verify users’ ages through government identification, but if Meta chooses to use a third-party verification service, that service can have a maximum false-positive rate of only 10 percent. If I were to bet, Meta will do its own verification.

It is no surprise that I think Meta’s business model is unethical, but ultimately, it is not illegal to design an addictive online platform. Neither is it illegal to mislead consumers into thinking that platform is safe for children. It is not illegal under U.S. law to send teenagers push notifications during school hours. The states’ settlement with Meta centers around ethics after the landmark ruling in March, when a New Mexico jury ruled Meta’s platforms were marketed and designed unethically. The state’s consumer protection laws allow the attorney general to sue platforms violating a narrow set of rules, but that does not make such a violation illegal. That was a civil case.

I’m centering my discussion of this settlement around this point because I think it’s important. Yes, Meta should be held accountable for the disaster it has intentionally unleashed on teenagers across the globe. Yes, Meta should be forced to change its platform design through negotiation and threats. Yes, Meta should be forced to pay damages for the unmistakable impact it has had on teenagers’ mental health in America. But no, Meta’s behavior is not illegal. It is not the responsibility of the state to nanny what content people consume, outside of obviously illicit or damaging media. Teenagers are people, guaranteed their First Amendment rights to use the internet without government permission. To that end, I find the results of this settlement disappointing.

I again will reiterate that I am broadly in favor of forcing Meta to redo its platform design. Every consumer should be given options to opt out of the algorithm, reduce nuisance notifications, hide like counts, and opt out of targeted advertising. These rights should be codified in U.S. law under comprehensive federal privacy regulation, which is long overdue. Meta has an unmistakable chokehold over online communication in the United States, and that should give the government the right to let Americans manage their experience on Meta platforms. But what we should not be doing is circumventing the law and applying state rules to a small part of the population that cannot even vote to express their displeasure. (To be clear, I am not advocating giving teenagers suffrage.)

One more thing: I can’t believe the attorneys general were so naïve as to give Meta a get-out-of-jail-free card if its competitors don’t comply with the same rules. Meta loves to make things about its competitors, and it loves crying wolf to the public. Remember App Tracking Transparency, an iOS feature so clearly designed to benefit consumers? Meta ran full-page ads in The New York Times about how Apple was killing small businesses and Meta was fighting for the underdogs. It’s the same playbook five years later: Meta is thinking of the children — how altruistic! — but says it would be so harmful to the children if the competition didn’t also follow the same rules. Stop bringing the children into this! Meta’s actual intention is that all of the advertisers who cater to teenagers will take their business to YouTube or Snapchat, where their users aren’t subject to a curfew. The assault against ATT was not about small businesses, and this new push isn’t about the children.

Apple Refreshes the Mac mini and Mac Studio at Even Higher Prices

Eshu Marneedi — Wed, 26 Aug 2026 02:57:03 -0400

John Voorhees, reporting for MacStories:

Today, Apple announced a new M6 chip that will debut in the Mac mini and an M5 Ultra that is coming to the Mac Studio. The company also announced an M5 Pro Mac mini and an M5 Max Mac Studio. Let’s take a look at the headline specs.

The headline feature of the M6 chip is that it’s Apple’s first 2-nanometer chip. The chip also includes “a larger 12-core CPU complex with the world’s fastest CPU core, a larger 12-core GPU with Neural Accelerators, a Dual 16-core Neural Engine, and up to 170GB/s of unified memory bandwidth.”

As for the M5 Ultra, which debuts in the Mac Studio, it sounds like a beast. According to Apple, the chip uses a next-generation version of its UltraFusion technology, resulting in a quad-die architecture for the first time in an Apple silicon chip. In addition, the M5 Ultra features an “up-to-36-core CPU and up-to-80-core GPU with a massive 1.2TB/s of unified memory bandwidth, 50 percent more than M3 Ultra.”

Macs featuring the new chips won’t be available immediately. You can preorder both online today for delivery on September 22 with the caveat that the M5 Ultra Mac Studio with 512GB of memory won’t be available until later in October.

Prices have risen too. The Mac mini models are both $100 more than the models they replace, while the M5 Ultra Mac Studio’s base configuration has risen $200. However, the M5 Max Mac Studio’s starting price hasn’t changed.

This is not the first time Apple has announced a new chip architecture ahead of the new iPhone: the M4 was oddly announced in the organic-LED iPad Pro in May 2024 before the A18 and A18 Pro came later that year. I do still find the timing of this announcement to be peculiar, however, because the headline iPhone event is in two weeks. These computers will, in fact, ship four days after the iPhones are expected to. The only reasonable explanation is that Apple didn’t want to clutter the iPhone event with Mac news and also for some reason couldn’t wait until October to announce these new machines, so they came now. But that’s not a satisfying reason. Apple already raised the prices of these machines on June 25, and it has done so again. This is effectively the third price increase the base-model Mac mini has received this year:

Late April: The Mac mini with M4, 16 gigabytes of storage, and 256 GB of memory, priced at $600, was discontinued.
June 25: That Mac mini returns at $800.
Tuesday: That Mac mini is refreshed with an M5 at $900.

This is also the second price increase in two months that the high-end Mac Studio has received — its base price has gone from $4,000 to $5,500 over the summer, a remarkable 37.5 percent increase. So why didn’t Apple do just one price increase and announce the refreshes alongside it? It’s not like I didn’t expect this oddness. From my piece in June:

Apple was, and maybe still is, heavily rumored to announce new high-end MacBooks Pro with organic-LED touchscreens in the fall. But it appears Apple has either gotten into a fight with Taiwan Semiconductor Manufacturing Company over pricing and/or determined it won’t be able to ship enough units at these updated prices for manufacturing them to be worth it. Either Apple fully delays the new MacBooks Pro to late 2027, which would be truly remarkable as it would be over an 18-month gap between revisions, or it will ship them with M5 Pro and M5 Max processors this fall anyway. It’s undeniable that the price increases — both on the supplier and consumer sides — will have some effect on product timelines in the short term.

My best guess is that even Apple doesn’t know what’s going on, and it just now has enough liberty in its supply chain amidst high demand to release new models. This entire situation is completely un-Apple-like: raising prices three times over a summer, launching new models two weeks before an event, and disturbing the October Mac release cadence. (These computers were due for October.) I would not bet a cent on there being an October press release for new MacBooks Pro — Apple itself probably doesn’t know if it’ll be able to release new computers later this year. Neither would I bet on the prices, configurations, and processor options those Mac laptops will have, if they even come at all. It’s safe to assume, however, that prices will increase further, much like Tuesday’s launch. This situation is a mess.

The M6 and M5 Ultra are the two new processors. The M6 is Apple’s first on Taiwan Semiconductor Manufacturing Company’s 2-nanometer process, and it’ll almost certainly be joined by the A20 Pro system-on-a-chip in September. (The A20 is due alongside the base-model iPhones in the spring.) I’m eager to hear about thermal performance and reliability, since TSMC’s initial 3-nm process was a disaster: the M3 and A17 Pro processors notoriously ran warm, and those devices had poor battery life. (I speak from experience; I still have an M3 Max MacBook Pro.) Coupled with the rumors that there may not even be an M6 Pro and M6 Max, I’d really like to know what’s different about the M6. The M6 is Apple’s first to have three core types: efficiency, performance, and super, adopting the revised naming scheme from last year.

The M5 Ultra also has an interesting story: It is Apple’s first “quad” processor, i.e., four Pro-level processors fused together. The M1 Ultra fused two M1 Max processors using UltraFusion, a high-speed interconnect Apple is again using in the M5 Ultra to combine two M5 Max chips. After the M1 Ultra’s launch — but before the disappointing Apple silicon Mac Pro refresh — it was widely rumored that Apple was working on combining two M1 Ultra processors to form an “M1 Extreme” to power the Mac Pro or iMac Pro. That project was eventually canceled, and the Mac Pro shipped only with an M2 Ultra before being discontinued in favor of the Mac Studio. The M5 Ultra is the spiritual successor to that project and is, without a doubt, the most capable Mac Apple has ever released. Top-end versions cost upward of $18,000 — and those models don’t even have 512 GB of memory.

Some takeaways as we head into iPhone season:

As I predicted earlier in the summer, Apple’s supply chain and product release cycles are fully disrupted. I would say the fate of the high-end MacBooks Pro is unknown. I would not bet on the 512 GB Mac Studio becoming available in October, either.
Desktop Macs are price-prohibitive for normal consumers, and I expect that trend to continue as the Mac laptops — all three — are refreshed over the next year. $900 for a 256 GB Mac mini is patently absurd. It has been a long time since I called a Mac overpriced, but most Macs on sale this fall are indeed overpriced. It is not a good time to want a new computer.
I wouldn’t rule out further price increases, supply shortages, and product timeline uncertainty. This feels to me like it’s only the beginning — and that it will serve as one of John Ternus’ most pressing challenges as incoming chief executive.

A correction was made on September 1, 2026, at 12:12 a.m.: An earlier version of this article stated that the M5 Ultra processor was four M5 Max chips fused together through UltraFusion. This is incorrect. The M5 Ultra’s graphics processing unit is four M5 Pro chips combined; it can be configured with up to 80 cores — each M5 Pro contributes 20. The central processing unit is only the equivalent of two M5 Pro / M5 Max chips for a maximum of 38 cores — each M5 Pro contributes 18. I wrote more about the chiplet architecture these processors use back in March.

There Really Isn’t Much to Say About the Pixel Phones This Year

Eshu Marneedi — Thu, 20 Aug 2026 23:28:15 -0400

David Imel, reporting for The Verge:

Google’s new HiLight notification LED on the Pixel 11 Pro is nearly useless. Out of the box, the only two things it can glow for are when the phone is face down and you’re interacting with Gemini, or when you get a call from a favorite contact. And even then, it can only glow one of five colors!

This is a massive wasted opportunity for a feature that used to be so beloved on older Android phones, so developers have wasted no time doing what Google won’t.

X user Dhananjay_Tech developed a new app called HiLight Studio, which lets you do practically anything you want with the HiLight LED array. Did you receive a new Slack notification that mentions a certain word? Make HiLight pulse in a rainbow array. Get a new text? Make HiLight spin in a wave of green. You can even control things like timing per cycle, brightness, intensity, and saturation.

I’ve neglected writing about the Google Pixels this week because I haven’t got much to say about them. Just like almost every other technology product on the market in the mid-2020s, they add some new generative artificial intelligence features to justify nudging the price up by about $100 amid a crippling memory shortage caused by those very artificial intelligence features. The new HiLight LED is about the most canonical example of this phenomenon. Google clearly wanted to add a flashy Gemini feature to this year’s flagship, so it added an LED to the back of the smartphone and gave it a clever name. Thank goodness for this developer for making something out of this.

The most heavily marketed AI feature on this line of Pixel phones is called Proactive Assistance. Here’s an excerpt from Imel’s fantastic review:

Perhaps the largest push in this direction is the Pixel 11’s new “Proactive Assistance” feature. This is a set of dynamic chips that appear when certain interactions happen on your phone, like a friend asking what your schedule looks like next week or if you booked that table at a restaurant yet.

For the most basic interactions, like checking availability or adding an event to my calendar, I found Proactive Assistance really helpful. I practically live inside my calendar, and text conversations about scheduling are common. When my friend Marc texted me asking when I was available next week, Gemini popped up and suggested checking my calendar for times. Outside of simple scenarios like that, though, Proactive Assistance gets less reliable. One of Google’s key examples is letting Gemini book a dinner reservation for you if you’re chatting about heading out to eat, but I had to receive a text with an extremely specific string of information to get it to work — and even then it only worked sometimes…

One promise of large language models is the ability to understand what you mean and remove the requirements for smart keywords, so it’s odd that explicit keywords are still necessary to get so much of Proactive Assistance to work. I can see what Google is trying to do with this and I can maybe see a future where it gets refined, but as it stands, it’s simply too inconsistent. Plus, I had to manually toggle on Proactive Assistance and add OpenTable to connected apps in settings, which I just don’t think most people will know to do. As my colleague Cameron Faulkner puts it, “You need to be proactive to get Proactive Assistance to work.”

Proactive Assistance sounds like a generative AI-powered version of Siri Suggestions, which, alas, haven’t been updated with Siri AI yet. Because they rely on Apple’s old machine learning frameworks, I’ve always found them to be almost useless. For years, I’ve crowned the Google Pixel the best Android phone on the market purely because of Google’s excellent Android skin, which some incorrectly call “stock” Android. (It isn’t “stock” — it’s Google’s well-done take on the Android operating system.) But some years, Google misses the mark with its software, and this seems like one of them. I’d expect a lot more from the company that has, for years, championed its smartphone as “the most intelligent.” Continuing from Imel’s review:

As is tradition, Google also introduced a bunch of new camera features this year, and perhaps the most prominent is Magic Capture. The feature is front and center in the camera app right next to the photo and video modes, which tells you something about how confident Google is in it. Start a Magic Capture, point it at what’s happening, and when you’re done recording, Gemini will do some processing and pull out what it considers the best photos and video clips from the session.

This is one of the least objectionable generative AI-powered photography features Google has added to its smartphones in a while — consider me pleasantly surprised. I’m always torn between taking a ton of photos of important events — which I know I’ll look at later — and capturing one long video, which feels more true to form, albeit harder to skim. Magic Capture allows people to do both, all automatically, and I think it’s quite delightful. I was initially anxious about the “processing,” but it really seems quite minor, just enough to eliminate motion blur when it really must be avoided to preserve the shot. I feel like Google — and other technology companies embracing generative AI — would enjoy much more success in the public eye by building features like this, not ones that aim to redefine photography altogether.

I Don’t Want AirPods With Cameras

Eshu Marneedi — Wed, 19 Aug 2026 16:35:07 -0400

Juli Clover, reporting for MacRumors:

Apple is working on camera-equipped AirPods that appear to be nearly ready to launch, based on a video MacRumors found in the macOS Tahoe 26.7 release candidate.

In a short demo, a man holds a book up so the camera in the AirPods can see the title. “With Visual Intelligence, your world becomes savable. See something you like? Just ask me to save it for later,” says the voiceover text.

I have been on the record as saying I have no interest in this product. I love my AirPods Pro 3 and have them in my ears almost all the time; I find the rumors about this product also carrying the AirPods Pro name concerning. It will be unacceptable to me if I have to purchase these AirPods just to get the best sound quality and active noise cancellation performance. My hopeful guess is that the standard AirPods Pro 3 will remain for sale at $250, whereas these new ones will be marketed as “AirPods Pro 3 with Visual Intelligence” for, say, $350. (I’m hoping the two products are otherwise identical.) There are two main reasons why I don’t want camera-equipped AirPods: practicality and privacy.

First, practicality. From the leaked video — which is remarkable by itself; this is one of the most embarrassing Apple leaks in a while — the AirPods’ stems are noticeably thicker. I wear mine to sleep every night, and even AirPods Pro 3 protrude too far outward to be completely comfortable. AirPods should always be light and as minimal as possible. I also just don’t find a practical reason for putting cameras in AirPods to begin with. Perhaps this is just me, but I have never picked up a book and wanted to save it. I have never held some nuts in my hand and asked myself how much protein is in them. I don’t go up to random people’s dogs and ask Visual Intelligence what breed they are. I’m yet to hear a single compelling use case for Visual Intelligence.

Second, privacy. Society is undergoing a mass revolution against surveillance. Flock license plate cameras, Meta Ray-Ban sunglasses, and artificial intelligence data scraping are hugely unpopular throughout the United States. I single out the United States because Americans are not averse to relinquishing their privacy for useful software — there has been no major push for data privacy regulation within the United States on a federal level, and Americans largely accept targeted advertising. But AI-enabled surveillance is widely unpopular politically and socially because Americans don’t see its utility, a position I largely agree with. There’s not much compelling about AI-enabled cameras everywhere. Camera-equipped AirPods will be symbolic of a Big Tech-powered surveillance state.

As a commentator, I’m intrigued to see how Apple spins this. But as a citizen and user of Apple products, I’m just not excited for yet another gadget with AI cameras.

Anthropic Plans to Watermark Claude-Generated Text

Eshu Marneedi — Sun, 16 Aug 2026 15:21:25 -0400

Anthropic:

Future Claude models will generate text that contains a watermark. This is a way of determining the likelihood that Claude was involved in writing the text, and we, along with several other major AI providers, are implementing this change to comply with the EU AI Act.

In this article, we share answers to some of the questions we’ve received about how our chosen watermarking method works, whether it affects Claude’s outputs, and why we’re making this change. To summarize:

We use a method of watermarking that does not have any practical impact on the quality or content of Claude’s outputs;

The difference between watermarked and un-watermarked text will not be distinguishable to readers;

Nothing is added to the text and there are no hidden characters;

Watermarking doesn’t require extra tokens, and will not be more expensive;

Watermarking carries no identifying information and can’t be traced to a specific person, organization, or chat;

Watermarking won’t be specific to Claude. As of August 2, the EU requires AI providers serving its market to mark AI-generated content. Other major model developers have signed the same Code of Practice and will be implementing their own watermarks.

Large language models like Claude work by generating one word at a time. Each time the model decides on the next word, it chooses among a list of potential candidates, ultimately selecting the most sensible or likely based on the preceding text. Take the sentence “The weather today was cold and…”. The next word is very unlikely to be “sugary.” But it is quite likely to be “overcast” or “grey.” Under most circumstances, it doesn’t matter much to the reader which of these latter two words the model ultimately chooses—the meaning of the sentence is largely the same either way. In cases like this, the choice is settled by a random number.

Watermarking uses low-stakes choices like these—which occur many times over a piece of generated text—to leave a pattern in Claude’s responses. That pattern is undetectable to the reader, but is detectable to anyone who has a key that encodes it. When watermarking is used, choices are still made at random, but the source of the randomness is different. Instead of using an arbitrary random number generator to pick the next word, watermarking uses the key and a few words that come before to settle what word the model should pick. That is, the words that Claude picks are still random, but now, one can check the sequence of words and see if it’s consistent with the choices Claude would make if it was using the key. If it is, one can assign a probability that the text was generated by Claude.

Because the watermark is created through subtle differences in word choice, it won’t apply to code, which requires exact syntax to function properly.

I think it’s important for there to be a foolproof way to identify artificial intelligence-generated outputs, whether they be images, text, or audio. Social media platforms should also prominently mark AI-generated images and text as synthetic, especially if the content is political or official. I stipulate that identifying AI-generated content should be foolproof because false positives can have detrimental consequences. Pangram, the leading AI text detector which doesn’t — at least for now — rely on watermarking, is widely regarded as the most accurate but still can occasionally flag human-created text as AI-generated. Universities across the globe punish their students for passing off AI-generated work as their own; the consequences of these systems are enormous.

SynthID, a framework developed by Google DeepMind, embeds an invisible watermark within AI-generated images and videos. It remains intact even after manipulation or screenshots, and is completely random, meaning it cannot be faked or removed. SynthID is used by Google’s Nano Banana image generator and ChatGPT’s image generation feature, two of the most popular tools. Social media websites like Instagram, Facebook, X, and YouTube scan uploaded content for SynthID watermarks and display a badge next to synthetic imagery. SynthID is foolproof and one of the best systems for detecting AI-generated imagery online, but it just can’t be replicated via text.

This new watermarking framework outlined Friday is Anthropic’s first try at developing a text watermarking technology. The problem is that it objectively isn’t foolproof — again, SynthID cannot be replicated with text. I’m less concerned that the outputs will be lower in quality than those without the watermark, but I am worried about false positives. Text that has been edited by Claude shouldn’t have a watermark at all, whereas Claude-generated text edited by a human absolutely should. Here’s what Anthropic says:

Light editing probably won’t remove the watermark completely; a complete rewrite where every word is replaced will. In the latter case, of course, it’s arguable whether the text can any longer be described as AI-generated…

The watermark only applies to words Claude chooses. When Claude proofreads text written by a person, what it gives back has generally only been lightly edited; because nearly all the words are the person’s, there’s very little (if anything) for the watermark to attach to. Depending on the length of the text and how heavily Claude has edited it, those changes might not be enough to make Claude’s involvement detectable. The more Claude writes, the more decisions it has to make, and the more space there is for a watermark.

These paragraphs don’t project much certainty. “Probably won’t,” “generally only,” “might not be enough.” I don’t blame Anthropic for this — watermarking text is nearly impossible! Claude relies solely on word choice to watermark its outputs, and word choices can be changed trivially. But I’d argue that releasing a first-party watermarking tool that isn’t 100 percent accurate is a dangerous precedent. People will use this tool to check for AI-generated text, but because it’s a first-party framework, arguing against its judgment will be an uphill battle. This is not Pangram; it’s not a guess.

Again, I do think all AI-generated text in the future ought to be watermarked, but with the important caveat that the watermark is foolproof. This watermark doesn’t guarantee that, and I think it ought to be scrapped.

More on the Apple v. OpenAI Lawsuit

Eshu Marneedi — Wed, 05 Aug 2026 04:31:12 -0400

Marcus Mendes, reporting for 9to5Mac:

Which brings us to The Information’s report, and its account of several cases where “Apple employees who left for other companies over the past decade told The Information they, too, continued to have access to confidential documents after leaving the company, even though they made no effort to do so.”

From the report:

Former Apple employees say an unusual company policy that encourages them to blend professional and workplace technologies is why they unexpectedly ended up with access to confidential files after their departures.

When new employees join Apple, the company often issues them an iPhone and Mac and pays for an iCloud account with a large amount of online storage capacity. Crucially, during the onboarding process, Apple encourages new hires to use their preexisting personal Apple IDs with this iCloud account, through which their co-workers can share internal Apple documents and other files with them.

The reason behind this, The Information notes, is the fact that iPhone users “can only log into a single primary Apple ID that unlocks all iCloud capabilities at a time.” So to avoid carrying two different sets of devices, one for personal and another for work (which is not uncommon by any means), “most Apple employees opt to use their personal Apple IDs to access their iCloud accounts.”

The report explains that, since the late 2010s, Apple employees have had access to a company-managed folder within iCloud that is automatically removed when they leave.

However, many internal documents are reportedly not automatically saved there, allowing files shared outside the folder to remain mixed with employees’ personal data and accessible after their departure.

This iCloud news definitely makes the case look less like a slam dunk at first glance, but it’s still obviously winnable¹. Cases like these hinge on the idea that the plaintiff must take reasonable action to ensure its intellectual property stays confidential. This combined work and personal iCloud account structure will clearly be at the center of OpenAI’s argument, along with supposed evidence that it was actually Apple employees who messaged Chang Liu, the employee sued alongside OpenAI, for help with Apple internal projects after his departure. (OpenAI also claims, with seemingly no attached receipts, that Liu was trying to “do the right thing,” with no further elaboration.) The idea that Apple doesn’t care about guarding its intellectual property is laughable — I don’t think it’ll work in court.

The messages between Liu and an Apple employee don’t refute Apple’s argument that OpenAI instructed and encouraged its Apple hires to bring prototypes and other confidential information into interviews. OpenAI’s Monday-evening response to the lawsuit only says, “Tang has always been clear with the team that we do not want, and must not use, any confidential information… Tang served Apple for more than 24 years and was widely known as one of the most innovative leaders at the company,” referring to Tang Tan, now OpenAI’s chief hardware officer. OpenAI also doesn’t address the assertion that it asked an overseas supplier for information about Apple’s confidential manufacturing processes. The blog post only truly refutes — with evidence — one claim relevant to the lawsuit: that Apple’s lawyers tried but failed to contact OpenAI’s. I’m not sure whether that will even matter in court.

I still think the case is winnable for Apple because even if the documents weren’t secured — either in iCloud or, as Apple says, third-party cloud storage — Apple could (correctly) assert that it wasn’t suing because the employees simply had access to documents post-employment, but that they actively used these documents to get a job at OpenAI. And, to involve the company itself, that using those accidentally obtained documents was encouraged by OpenAI’s chief recruiters and officers. It will argue that Apple employees knew they should report such incidents rather than take advantage of them for future employment. The motives of OpenAI’s executives and employees are central to the lawsuit and, if it goes to court, will have to be investigated during discovery. (It’s very much possible this settles out of court.)

An update was made on August 5, 2026, at 5:19 a.m.: I’ve added further clarification on the iCloud storage claim.

Notably, Apple asserts in its lawsuit that the claims it presents are related to “Apple’s third-party cloud storage,” not iCloud. (I have a feeling Apple knew this news about combined iCloud accounts would leak.) It’s still obvious that OpenAI will attempt to use the iCloud situation as evidence for its legal strategy — poking an inconsequential hole in Apple’s argument. It will not work. ↩︎

The Compromises of Folding Phones

Eshu Marneedi — Fri, 31 Jul 2026 17:10:56 -0400

Dominic Preston, reporting for The Verge earlier in July:

A year after overhauling its Z Fold phone with a radically thinner design, Samsung has changed things up again, this time altering the shape entirely. The Z Fold 8 is wider than any of Samsung’s previous foldables, and shorter too, a size and shape somewhere between Samsung’s traditional Flip and Fold models.

In photos, I wasn’t impressed by the squat shape. In person, to my surprise, it feels great, and almost immediately intuitive. If you want to get a feel for it yourself, grab your passport — the Z Fold 8 is about the same height and less than half an inch more narrow.

This isn’t Samsung’s only new foldable this year, and in fact, its lineup just got more confusing. The Z Fold 8 may be the official follow-up to last year’s impressive Z Fold 7, but in truth the Z Fold 8 Ultra is the straightforward sequel. That phone has the same squarish inner screen as previous Z Fold phones and a similar triple rear camera. The name feels like a sly trick to get around this year’s inevitable price hikes: Samsung slapped “Ultra” on the box of its highest-end model to help cover for pricing the phone at $2,099.99, up $100 from the Z Fold 7.

I don’t have much interest in Samsung phones but nevertheless have been linking to Samsung Unpacked news for years now. What’s more intriguing to me is this new form factor, which is widely speculated to be the same as that of the folding iPhone due in about six weeks. The first generation of folding smartphones, beginning with the Galaxy Fold in 2019, had long, narrow front screens and square inner screens. This aspect ratio made both screens less than ideal to use: the front screen was too narrow, and the inner screen wasn’t wide enough to be used as a tablet. This new “squat” form factor sacrifices vertical length for width, making the inner screen much more useful while sacrificing day-to-day use of the front screen.

I think this is the way to go, and I remember saying so when the Pixel Fold was first announced in 2023. The entire point of a folding phone is the inner display, which should be large enough to eliminate the need for a tablet. But this great aspect ratio debate of the 2020s has proven something more important: folding phones are inherently full of compromises. The front screen — which most people will use most of the time — must be sacrificed to make the inner screen useful. This is still better than the old design, where both screens weren’t ideal to use, but it’s still a compromise. For the best experience, people will open up the phone like a book. This style of folding phones ought to be called “tablets with front screens as an afterthought.”

This brings me to the foldable iPhone. I’m positive the software will be world-class. App developers will be there Day 1 to ensure their apps work well on both displays; Apple will make the front display useful and practical. But this device is rumored to cost upward of $2,000, while forgoing the professional-grade cameras of the iPhones Pro, the pocketable yet useful display size of the standard iPhones, and the battery life of the iPhones Pro Max. (Just compare the specifications of the Galaxy Z Fold 8 with the Galaxy S26 Ultra.) I’m almost positive I won’t want to carry two displays around in my pocket, especially because I’ll have to open the phone up when I truly want a good experience. I’m almost positive I don’t want to spend $2,000 for worse cameras and battery life. And I’m almost positive I’m in the overwhelming majority.

I’ll refrain from passing judgment on the foldable iPhone until it’s announced, but I can judge the Galaxy Z Fold 8 — and I think it exposes too many compromises with this device class. None of this is Samsung’s fault, of course, but it does a great job illustrating why foldable phones will probably never be ready for primetime. There was a time many years ago when I was bullish on the folding phone, hoping technology would progress far enough that they’d be equivalent in price and features to their non-folding counterparts. That time has not come yet, and I’m not confident it ever will. Samsung and Apple have tried addressing the biggest drawback to folding phones, the form factor, this year. But the Galaxy Z Fold 8 proves that it’s just not enough.

‘You’ll Own Nothing And You’ll Be Happy’

Eshu Marneedi — Tue, 28 Jul 2026 16:50:01 -0400

Chance Miller, reporting for 9to5Mac:

Apple has officially announced Apple Upgrade, a new product leasing program available starting today in the United States.

Apple Upgrade, which is backed by Klarna, lets users lease iPhone, Apple Watch, Mac, and iPad purchases, with the option to easily upgrade at the end of the lease term. Here’s how it works.

Apple Upgrade is available across Apple’s product lines, including the latest iPhone 17 lineup, MacBook Air, Apple Watch Series 11, and much more.

iPhone and Apple Watch are available with 12-month and 24-month leasing terms, while Mac and iPad are available with 24-month and 36-month terms.

iPhone leases start at $17.99 per month for iPhone 17e

Apple Watch leases start at $11.99 per month for Apple Watch Series 11

Mac leases start at $24.99 per month for MacBook Air

iPad leases start at $15.99 per month for iPad mini

On the surface, this looks like a great deal. People can now lease any Apple product — including Macs with custom specifications — and upgrade their device at the end of the lease. They can also buy it, but then they’ll have to pay the remaining cost: Leasing a $2,000 MacBook Pro for 36 months only costs $39 a month, meaning someone would have to pay nearly $600 at the end of their term to own the product. At that point, you might as well just upgrade at no cost. These leases accrue no interest, either, so it seems like a win to lease a product like this and invest the money somewhere. That’s the pro-Apple Upgrade argument, at least.

I don’t like these leasing programs. I never used the now-discontinued iPhone Upgrade Program, despite buying a new iPhone every year, and this is the same reason why I encourage everyone I know to never purchase an iPhone through a carrier contract, even if they’re “handing them out for free.” There’s a monetary aspect to this — people living above their means; this program will make some people likely to purchase a device they can’t afford and become delinquent on their payments — but also a societal one. Why can’t we own anything anymore? Why is it that houses are so expensive that people are better off renting in most markets? Why are video games going digital-only, with software licenses that stipulate that you don’t actually own the games?

There are undoubtedly people who will benefit from this program. If Apple Upgrade lets you buy a computer that kickstarts your career, by all means, enjoy it. (Just please don’t buy a computer you couldn’t afford if push comes to shove.) But I take issue with not being able to own my smartphone — I don’t want a tool that only costs a thousand dollars to be tied to a lender who could theoretically take it away at any moment. That’s a dangerous precedent; at what point will everything we know and love be reduced to installment payments? Someone somewhere must draw the line. We have leases on cars because they cost tens of thousands of dollars — few people have that much in cash. But have we made it so far south in society that people need to lease computers?

This program rubs me the wrong way. Maybe it doesn’t for you — and that’s completely fine. But if it doesn’t, maybe you really will own nothing and be happy.

Apple Launches the MapKit for Automotive SDK, First With Ford

Eshu Marneedi — Fri, 24 Jul 2026 04:14:44 -0400

Apple Newsroom:

Apple and Ford today announced that Apple Maps will be integrated directly into Ford’s upcoming Universal Electric Vehicle (UEV) Platform through Apple’s new MapKit for Automotive SDK. Coming to Ford’s UEV Platform in 2027, this integration will deliver a beautiful and easy-to-use navigation experience powered by Apple Maps directly to the vehicle’s displays. Road-level Maps information will also enable Ford’s Latitude AI team to build a seamless hands-free driving experience.

Leveraging Apple’s new MapKit for Automotive SDK, Ford’s Universal Electric Vehicle Platform will offer drivers turn-by-turn directions using natural language, real-time traffic and incident information, intuitive search featuring detailed place cards, and routing options to help users easily find the best route to their destination. Additionally, Ford UEV drivers will benefit from an intelligent EV routing functionality, including battery preconditioning. Ford will also use road-level information from Apple Maps as part of the development of its next-generation BlueCruise hands-free highway driving capability to deliver a more seamless on-ramp-to-off-ramp experience.

Apple services have historically been sold as complements to Apple hardware: Apple Maps, for a decade, had been limited to Apple devices, replacing Google Maps. That is no longer the case; Apple TV is a canonical example of Apple’s interest in broadening the Apple services market. Similarly, this new MapKit SDK — short for software development kit — is an intentional expansion of Apple’s services market following the abject failure of CarPlay Ultra, a version of CarPlay meant to consume the entire car software experience. Even CarPlay’s most fervent supporters, like BMW and Ford, haven’t adopted CarPlay Ultra because it exerts too much control over a driver’s experience. The CarPlay Ultra gambit was clearly an effort to make something out of the Apple Car project, also a failure.

Ultimately, I think the CarPlay Ultra fiasco is indicative of the two opposing yet symbiotic forces within Apple: the product people and the services people. The product people want Apple to own as much of the “stack” as possible: the hardware, the software that hardware runs, and the services that software sells. They wanted to build a car, make software for it, and sell services through that software: that was Project Titan, the car project, in a nutshell. That is fundamentally the same strategy that governs the Apple Watch, MacBook Neo, even the iPhone. The services people, like those in charge of Apple TV and its content acquisitions like Formula 1 and Major League Soccer, don’t believe in that because they’d rather make money through advertisements and subscriptions. They paid hundreds of millions of dollars in upfront costs for MLS and F1 to sell more subscriptions, even to people who have never owned an Apple device.

The services people won here, largely because the product-driven CarPlay Ultra / Project Titan strategy — to own as much of the software experience as possible — failed. No car manufacturer was interested in that. Perhaps consumers would’ve been, since CarPlay by itself is overwhelmingly popular by Apple’s numbers, but it didn’t matter anyway because the difference in experience for iPhone users and non-iPhone users was too stark. (Not to mention, it was too much work for very little tangible benefit on the manufacturers’ end.) So now, Apple Maps — just like Apple Music in Tesla, General Motors, and Rivian vehicles — is a service built to make money independently. It doesn’t matter whether someone owns an iPhone or not, because they’ll still be forced to use Apple Maps in their Ford vehicle. And most importantly, they’ll be forced to look at Apple’s advertisements coming to Apple Maps soon.

The ads part is a tragedy because it signals a definite end to Apple’s product-driven mindset. Before Apple started putting ads in Apple services, they were also Apple products — you buy a subscription to Apple Music. Now, Apple services are partially ad-driven enterprises, where your use of the service itself draws revenue. In creating the MapKit for Automotive SDK, Apple doesn’t care about people spending money at the Apple Store or on in-app purchases. Apple isn’t delivering a product to consumers as much as it’s guaranteeing engagement to ad buyers. The consumers are not the primary consumers; corporations who pay Apple are. This is against the very philosophy Apple in many ways was founded upon: making high-quality products for consumers willing to pay premium prices. Modern Apple would probably object to the characterization that ads in Apple services cheapen the experience, but it’s true: ads in Apple Maps do not make for a premium experience.

The new Apple doesn’t seem to care about that. It’s uninterested in delivering a premium experience, and it’s uninterested in asking for your money to deliver that experience. Ergo, the MapKit for Automotive SDK.

U.S. Threatens to Sanction China Over AI Distillation

Eshu Marneedi — Wed, 22 Jul 2026 00:22:34 -0400

Ashley Capoot, reporting for CNBC:

U.S. Treasury Secretary Scott Bessent on Tuesday said the Trump administration will look into whether Chinese artificial intelligence models have been distilled from American models, stating that the government does not support “IP theft.”

Chinese open-weight models are gaining steam against leading offerings from American companies like OpenAI and Anthropic, sparking concerns from tech executives and government officials about the durability of the U.S. lead in the AI race. Moonshot AI, a Chinese startup, released a model called Kimi K3 earlier this month that outperforms those companies across some industry benchmarks. Open weight refers to models whose final trained parameters are publicly released for download, while the underlying code and training data remain private.

“If we see, especially that overseas models are stealing from our great companies, we have the ability to sanction them because of this theft,” Bessent told Fox Business’ “Mornings with Maria” on Tuesday.

Bessent said the technical term for this theft is called distillation, which is an AI training method where a smaller, less capable model is built using outputs from an existing, stronger model. Anthropic sent a letter to the U.S. Senate Committee on Banking, Housing, and Urban Affairs last month alleging that the Chinese tech company Alibaba had carried out the “the largest known distillation attack” against it to date.

If distillation — training on data that doesn’t belong to the Chinese labs — is a sanctionable offense, then why isn’t pre-training? How is it fully legal for Chinese companies to use training data published by American citizens on the internet? This whole argument is bogus, and it’s quite evident that Bessent knows very little, if anything, about what he’s talking about. AI models are built on the idea that training a computer on the open internet is fair use — that pre-training is transformative enough under fair use doctrine, not “stealing.” American companies can train their models on Chinese text and vice versa because pre-training is inherently transformative, an argument I’m quite sympathetic to. I think that if we built a time machine back to 2022, a permission or compensation structure should’ve been created, but alas, this is the system we’re stuck with. The world’s copyright system is not suited for AI pre-training.

Back to the Bessent argument. Anthropic, and perhaps OpenAI, are perturbed that Chinese models — like Moonshot’s Kimi K3 open-weight large language model — were trained via distillation. Kimi K3 is cheaper than Claude Opus 4.8 and GPT-5.6 Sol, yet almost matches GPT-5.6 Sol in overall intelligence and beats Opus 4.8 entirely. Kimi K3 comes just shy of Claude Fable 5, a model over three times the price. While Anthropic can’t even be compelled to offer Fable 5 access to some of its paying customers, Kimi K3 offers similar performance at a fraction of the cost — but it was built by distilling Anthropic’s models. Anthropic’s view, which it has surely expressed to the government, is that this is an adversarial nation subsidizing AI development, stealing the secrets of American AI labs, and building a compelling product for much cheaper. From that perspective, Moonshot and China look guilty.

But that’s a pacifist view; Anthropic is wailing for pity from a government that has treated it abysmally. Just Tuesday, a judge approved a billion-dollar settlement between Anthropic and a litany of authors after it pirated books to train its models on — is that not “stealing?” Is Anthropic not one of the most blatant copyright infringers of the modern era? It’s quite obvious here that Anthropic is crying wolf not because China gained some unfair advantage in its training process, but because Anthropic is so upset that it wasted perhaps billions of dollars in subsidized tokens all for its users to prefer a foreign competitor’s clearly superior model. The problem here was never about morality or copyright or distillation; the AI labs are morally bankrupt, and they know that. If they didn’t, they wouldn’t be cutting billion-dollar checks to authors.

So where do we go from here? I think the solution is quite obvious: Anthropic must start distilling its own models. Train a massively expensive competitor to Fable 5, distill a new version of Opus on that model’s reasoning traces, and ship the new Opus at a much lower cost. We’re reaching a point in the AI industry where cost is a massive factor in the models enterprises and individuals use, as reasoning traces become more expensive. Yes, pre-training is currently the most expensive part of releasing a new model, but we’re already seeing the shift toward inference and post-training. New model releases nowadays are not new pre-training runs; GPT-5, the last “major” AI model from OpenAI, was a heavily post-trained GPT-4o from spring 2024. As LLMs mature and reasoning tokens become more valuable, inference will be the biggest operating cost, not pre-training. Train one model and distill three others, then offer inference at a competitive price. Distillation — from China and by the American labs themselves — is in the AI industry’s best interests. It saves on pre-training costs and leaves more room to make healthy margins on inference.

Apple Is Looking Into Buying AI Chip Companies for Private Cloud Compute

Eshu Marneedi — Sun, 19 Jul 2026 01:50:02 -0400

Marcus Mendes, reporting for 9to5Mac:

According to The Information, Apple has been exploring potential acquisitions of semiconductor startups to “boost its efforts to build server chips for running AI.”

From the report:

In recent months, the iPhone maker has talked with bankers about possible deals. It has also approached semiconductor startups to gauge their interest in selling themselves, the people said. Apple’s hunt for chip acquisitions comes as the company struggles with the performance of its own internal AI servers, which currently run on internally designed M2 Ultra chips.

The Information says Apple’s next-generation server chip, code-named Baltra, has slipped past its planned 2026 debut. In the meantime, Apple is using M2 Ultra-based systems for some of its own AI processing, while the more demanding tasks are handled by the Gemini-based model powering the new Siri, which runs on Nvidia GPUs in Google Cloud.

A big acquisition would be a departure from Apple’s usual M.O. of buying smaller startups from time to time, but it wouldn’t be unprecedented.

It would indeed be unusual for Apple to purchase a Silicon Valley start-up specializing in artificial intelligence processors, when Apple arguably makes the best consumer AI chips. As Bloomberg has reported extensively, Apple’s consumer processor strategy — for high-end Mac models, at least — has shifted considerably toward AI. People like buying Mac Studios and Mac minis to do local AI inference — they’ve become popular as cloud computers and mini data centers. Per The Information’s reporting, Apple’s internal processors are also on that trajectory, albeit delayed for unclear reasons. Apple’s processors aren’t just fast and capable of AI inference, but they’re extremely efficient, which makes them relatively inexpensive to use in the long run. It’s no wonder why Apple has retaken the No. 1 market capitalization, surpassing Nvidia on Friday.

Regardless, this news is quite interesting because it proves Apple is intent on shipping more powerful, capable models in the future — perhaps ones that can write code or perform agentic tasks. It wants people to be able to run those models on their own devices — hence why the company is testing Mac Studios with 1.5 terabytes of memory and skipping high-end M6-generation chips entirely — and to eliminate its reliance on Google Cloud, or maybe even Google altogether. If Apple were to buy such a company, it could probably pre-train its own models: a lack of compute was one of the key limitations that led to the Google partnership. Apple’s new leadership doesn’t see AI as a fleeting trend. It is finally taking it seriously, something we couldn’t say about the AI team under John Giannandrea, the machine learning chief who was stripped of his responsibilities and replaced by Mike Rockwell, the executive previously in charge of Apple Vision Pro.

I think Apple should lean into efficient, low-cost AI inference on both the hardware and software fronts. Siri AI clearly emphasizes on-device processing, and the company’s future Apple silicon plans do, too. Readers of this blog know that my single biggest qualm with the AI industry for now is that its economics aren’t rooted in logic. Anthropic is compute-constrained, so much so that they’re under heavy competition from Chinese AI labs that can offer inexpensive tokens and relatively the same performance as Anthropic’s best models. OpenAI is deathly unprofitable, leading some pundits to believe that its economics will spell its demise. Compute is the biggest story of Silicon Valley this summer, whether it’s skyrocketing memory prices for consumer products or unusual deals like Meta selling compute to Anthropic.

Whatever happens here — if Apple chooses to buy a compute start-up or not — the fact that it is even entertaining the idea is proof that it sees Apple silicon as its golden ticket in the AI space. As open-weight models get smaller and more powerful, people will buy more powerful Macs marketed as the best for AI inference; as the frontier AI labs become more compute constrained — and their flawed economics catch up to them — people will look to Apple’s models and products. Siri AI is merely the beginning of the Apple AI story.

A Bluetooth Smart Speaker Will Not Fix OpenAI’s Problems

Eshu Marneedi — Thu, 16 Jul 2026 01:16:54 -0400

Mark Gurman, reporting for Bloomberg:

OpenAI’s much-anticipated push into consumer devices is slated to begin with a mobile, screen-free smart speaker designed to be a new type of home computer for the AI era, according to people familiar with the matter.

The product — still under development — is meant to serve as a humanlike AI companion that lives in the home, said the people, who asked not to be identified because the project hasn’t been announced. It will help control smart-home appliances, play media, answer questions, respond to messages and tap into the range of capabilities offered by OpenAI’s ChatGPT, they said.

Though the new product resembles a speaker, OpenAI internally describes it as the first of its kind: a computer built for AI to help make busy people more productive. It includes a camera and other sensors that help it understand a user’s surroundings and context, as well as advanced AI models beyond those available on conventional smart speakers.

Remember Jibo? Jibo, developed by a professor at the Massachusetts Institute of Technology after a successful Kickstarter campaign, was an animated voice robot that would move around with a motorized base and use its camera to detect facial expressions and gestures. Jibo came years before the modern large language model, and its use was limited, but I’d imagine this device from OpenAI to be much like it. Like Jibo, this is just not a breakthrough consumer product. It’s a Bluetooth smart speaker with a motorized base — which Gurman describes as “mechanical elements that can move on their own” — and ChatGPT built in.

You can already buy a smart speaker with an LLM for just $100: it’s the Google Home Speaker, and — news flash — it’s not a revolutionary product. It’s certainly not revolutionary enough to warrant buying Jony Ive’s hardware company and poaching trade secrets from Apple. But none of this even considers how unfeasible it will be to have a hardware product from OpenAI — a company almost universally taken with distrust — listening and peering into people’s homes all day long. I’d be hard-pressed to find a single person who wants that. There are people on the streets irate about OpenAI data centers being built in their backyards; on what planet will those people put an OpenAI product in their homes?

OpenAI has spent billions of dollars on this hardware venture. It has upended the technology industry — it has decimated entry-level jobs, transformed computer science education, and sowed doubt in tech as a whole. People are down on artificial intelligence and technology writ large. For this product — and OpenAI as a company — to be successful, it must transform not only its public image, but that of the technology it wants to sell. I have a hard time believing a Bluetooth smart speaker designed by Ive with ChatGPT built in is a serious solution to that problem. Maybe it’s the most beautiful speaker in the world. Maybe it’s made with gold or some opulent frills. That doesn’t change the fact that OpenAI has a gargantuan favorability problem.

OpenAI also has a money problem. People do not like paying for AI. Only a small fraction of ChatGPT users are paying subscribers, and most of those people are on the lowest subscription tiers. The people who do pay heavily for AI burn subsidized tokens so hard that Sam Altman, the company’s chief executive, is holding back on an initial public offering until the company’s finances are more stable. What company that is struggling to sell consumers on a $20 subscription and enterprises at base rates for its models could possibly think it could sell hardware? The one popular product OpenAI makes, ChatGPT, is not profitable. It can’t even persuade people to pay for its flagship innovation, let alone a designer smart speaker.

I really don’t want to be Steve Ballmer. “$500? Fully subsidized? With a plan?” But I have a feeling this take will fare much better than my take from last year. Over a year later, Apple doesn’t look bad anymore; OpenAI does.

Hands-on With iOS and iPadOS 27, macOS 27 Golden Gate, and Siri AI

Eshu Marneedi — Mon, 13 Jul 2026 16:51:06 -0400

A return to form and function

Image: Apple.

In 2009, at Apple’s Worldwide Developers Conference, Bertrand Serlet walked onstage and presented an audacious claim: “Zero New Features,” the slide read.

Serlet, Apple’s then-senior vice president of software engineering, was introducing OS X 10.6 Snow Leopard, a release Apple claimed would emphasize small “refinements” and major updates to internal technologies. But to the public, it really was “Zero New Features” — a quote forever cemented in Apple’s modern history as one of its most notable. As Serlet later explained, OS X Snow Leopard had plenty of new features, most notably Grand Central Dispatch, which brought multithreading to OS X for the first time. OS X Snow Leopard established the architecture on which all personal computers, including the iPhone and the Apple Watch, would eventually rely. Its stability solidified Apple’s reputation for reliable, high-quality software, so much so that the coming decade of pervasive, sloppily thought-out user interfaces — beginning with OS X 10.10 Yosemite — was treated as an aberration.

In early June, Apple’s first real announcement at WWDC was neither novel nor revolutionary. It was not a coding agent that would put all white-collar workers out of jobs by 2030, augmented reality glasses, or even as much as a tab bar in the Photos app — it was a slider to adjust the translucency of Liquid Glass. Then came the stunning admission that macOS 26 Tahoe’s window corner radii were too inconsistent and radical: Apple would decrease and equalize all corner radii throughout the system in macOS 27 Golden Gate. Liquid Glass app icons would use new glass layers for a more polished, legible appearance. The central processing unit scheduler — the program that tells the CPU when to execute a process — was made more efficient. The Spotlight index was rewritten, and the Mail app’s search function was markedly improved. These were among the first “features” on which Apple spent the precious opening minutes of one of its most-watched presentations, and it was marvelous.

All of Apple’s releases this year converge on a single philosophy: returning Apple to its original promise of reliable yet innovative software. And much as OS X Snow Leopard — the company’s last true “cleanup year,” with hundreds of bug fixes and underlying performance improvements — became known for transforming personal computing for decades to come, so will iOS 27 and macOS Golden Gate. After a section of the keynote aimed, candidly, at regulators keen to scrutinize Apple’s child safety efforts, Apple introduced what I’d loosely call the Grand Central Dispatch of WWDC 2026: Siri AI, a completely reworked, large language model-powered version of Siri promised two years ago. After years of rumors, personnel changes, and bottlenecks, the “more personalized Siri” is finally here, and it’s both a Herculean reworking of the internals and a breakthrough consumer product.

I have spent over a month with all the new operating systems and Siri AI, and I can confidently report that Apple Intelligence, or at least this new version, has changed the way I use Apple products. It is the first artificial intelligence system that feels personal, like it knows me and is willing to use the decades of information I have stored on my Mac and iPhone. It sheds the gimmickry of current AI systems and, dare I say, makes you feel like you’re living in the future. Apple has proven that the answer to modern computing’s biggest questions comes not via the LLM alone, but via the cowling built around the model. Apple is a product company, and Siri AI is the best LLM product, engineered by people who astutely know what makes a great one. It’s an ice-cold glass of water in the hell of coding agents that will supposedly assume everyone’s jobs by the decade’s end.

But the “Snow Leopard” nature of this year’s operating systems leaves little to write about. There is, of course, Siri AI, coupled with some interesting changes to Liquid Glass, but there are almost no new features in this year’s releases. That’s OK — more than tolerable, in my book. I can ostensibly name dozens of features Apple announced in just a few years that occupied collective hours of keynote time, but hardly anyone uses: StandBy, Image Playground, tinted app icons, the Journal app, and Stage Manager, as pertinent examples. It’s not even that these features are useless, per se, but that they’re just too onerous to find. This was, in fact, the lede of my OS review from two years ago: Apple adds dozens of new, marquee features each year to go undiscovered and unused, while the core parts of the OS languish. Perhaps hundreds of millions of people search for emails in the Mail app, and it has been terrible for years. It isn’t anymore.

This is to say that my relative brevity in this year’s review should not be considered a gripe, but great praise. The adage of “comparison is the thief of joy,” begone — this year’s releases are a perfect replica of what made OS X Snow Leopard one of the most beloved Apple platform releases in history. They’re reliable yet teetering-on-the-edge-of-revolutionary pieces of software, and they’ll undoubtedly change the way millions of people use their devices in the fall. And I think a throwback to 2009 — a mélange between innovation and reliability — is the ideal place to be in a time of grave uncertainty and mistrust in the technology industry.

Siri AI

Siri AI is Apple’s marketing name for this year’s preeminent feature: a generative artificial intelligence-powered version of Siri that can search the web, perform actions in apps through App Intents, and peer into the personal context — a collection of personal data from first- and third-party apps. At its core, Siri AI uses a suite of on-device and cloud models to accomplish these three primary objectives. For more detail on how these models work in tandem, read my initial reactions to the Siri AI announcement from June. An on-device model, called the system orchestrator, chooses which model is best suited for the query and what information it needs, such as the personal context, an App Intent, or a web search. The model then uses the required tools to do the requested work and returns an answer.

This admittedly oversimplified description may be reminiscent of a chatbot, like ChatGPT, Claude, or — especially — Gemini. But over my weeks with Siri AI, I think the product is far closer to an interactive assistant than to a chatbot. Products like ChatGPT have taught us over nearly four years that natural-language input is the future of computation. Older virtual assistants, like the non-AI Siri and Google Assistant, used a limited understanding of human language to provide deterministic, reliable answers. They could neither process language naturally nor generate language of their own; they were instead hand-trained to recognize certain phrases like “What’s the weather tomorrow?” They were handed a predefined set of tools and were essentially told to match certain queries to them. In Siri’s case, the assistant only generated preprogrammed outputs, handwritten by Apple employees for different scenarios.

All of this is to say that those assistants were only useful for a handful of tasks, usually ones performed on-device. Siri — to this day, in non-beta versions of the operating systems — types almost all knowledge-based queries into Bing. It has virtually no knowledge of its own. Chatbots lie on the opposite end of the virtual-assistant spectrum: They claim to hold the world’s knowledge and have such a grasp on it that they can write complex computer programs from scratch or solve problems that have stumped mathematicians for decades. They’re mathematical models of human language, designed to predict the answer to a question nondeterministically. Siri AI lands somewhere in the middle of the spectrum, still unquestionably best suited, like its prior versions, for personal commands, yet with the breadth of knowledge that inherently comes from an LLM foundation. Siri AI merely leverages an LLM to realize the original product’s vision from 2011 — it is unapologetically not a chatbot, despite using the same underlying technology.

I say this for two reasons: the models and the interface. The former is straightforward: The models are nothing like GPT-5.6 Sol from OpenAI or Claude Fable 5 from Anthropic. AI enthusiasts use terms like “big model smell” to refer to the breadth of knowledge these frontier models have, and it’s immediately apparent upon interacting with Siri AI that “big model smell” was never Apple’s intention. Siri AI speaks organically, in short sentences, and uses little formatting. It prefers to quote much of the source material, such as when searching the personal context or the web. The model seldom launches into long reasoning traces and usually generates laconic responses. It’s lightning-fast — almost latency-free, to my delight — in nearly every interaction, choosing to process most queries on-device. Those are not characteristics of frontier-level models, which write with a certain cadence that Siri AI cannot match. Siri AI would clearly struggle to answer first-year calculus problems, let alone discover novel mathematics or write code; it is neither amusing nor particularly useful as a conversational partner.

The latter reason is, broadly, the rest of this section. People can open Siri AI on iOS in four ways: pressing and holding the Side Button to launch voice mode, using the Siri app, swiping down from the center of the Home Screen to open Spotlight, or swiping down from the top center of any app to launch the same Spotlight interface. The point is that Siri AI is available everywhere, including in apps. On macOS, people can use Spotlight or the “Hey Siri” wake word. Spotlight and Siri are not the same; rather, Spotlight leads to Siri. Spotlight works normally without any AI features unless the query no longer matches any results, in which case it offers to send it to Siri. If you prefer to talk to Siri directly, the voice mode bypasses Spotlight entirely, and so does the new, bespoke Siri app, which aggregates Siri conversations across devices through iCloud. Searching for “Siri” in Spotlight launches the Siri app.

Siri AI across modalities.

All of Siri AI’s new modalities use a beautiful semitransparent gradient; the traditional voice mode uses a new glassy orb. When using Siri AI from anywhere but the Siri app, it initially displays a short answer no longer than a paragraph. Manually swiping down on the initial result opens a mini window with a longer answer and, for select queries, web images. You can open the conversation in the Siri app from this window using a button at the top trailing edge. This reveals how Apple envisions Siri AI: It’s for direct answers, not long chats. People don’t even need to interact with the Siri app at all if they don’t want to — the app functions more as a place to recall past conversations than to start new chats. It is, of course, still possible to start chats and voice conversations from the Siri app — because anything else would be unintuitive — but that’s clearly not the app’s primary purpose. The Siri app is not a competitor to ChatGPT and isn’t a place where people will spend significant time.

The Siri AI app in iOS 27.

In practice, I start almost all of my Siri conversations on iOS through voice mode¹ — when I’m not in public, at least — and on macOS through Spotlight, just because speaking to a desktop computer seems daft. I scarcely even open the Siri app because it’s quite barebones, at least for now. It is neither as visually appealing nor as useful as the ChatGPT app, and offers only basic file uploads and search. I especially love all of the new animations and sounds when Siri is working: a circle of dots rotates as Siri reasons through an answer, and results drop down gracefully from the Dynamic Island on iOS. Results don’t stream in — Siri displays them in full only after it finishes writing the response. Siri AI is such a gorgeous, whimsical interface — it’s the canonical example of how best to use Liquid Glass. I also seldom go back and forth with Siri, except for clarifying questions, unlike my use of ChatGPT or Claude. Siri doesn’t invite conversation and feels more like a one-and-done assistant than a robot worth talking to or reasoning with.

The interface of Siri AI.

My most frequent use case for Siri AI is simple, fact-based questions. The model does occasionally hallucinate², and I wouldn’t use it for any serious work — such as fact-checking an article or learning a new topic — but it’s great for off-the-cuff questions. Some recent questions I’ve asked Siri AI include:

“When did Delta Air Lines take delivery of its first Airbus A350?”
“How many people voted for Kamala Harris in 2024?”
“Has India ever qualified for the World Cup?”
“Who won the Grammy Award for Best Alternative Music Album in 2024?” And then, “What about 2023?”

Siri answered each of these questions accurately, providing sufficient and reliable references for every one. These are all questions the older Siri would either refuse to answer — kicking off a Bing search instead — or send to ChatGPT, which would respond with varying degrees of success. The ChatGPT integration was also one-way: I could not follow up within the Siri interface, and chats started within Siri would usually take a few hours to appear in the ChatGPT app. And sure, I could use the ChatGPT app for these questions, but Siri is instant, almost to an uncanny degree. It writes a sentence or two, occasionally shows an image, and links to the site at the bottom of the response for easy access. Siri AI is simply better than ChatGPT for trivial general-knowledge questions, thanks to the way it presents information.

This is not to say that Siri AI never makes mistakes. The fast on-device models Siri AI opts for most are woefully out of date and, frankly, often incompetent. It all comes down to the basic premise I outlined at the beginning of this section: Siri AI is not a chatbot. I have worked through many projects with Claude and ChatGPT, but Siri AI was unable to pick up on any of them. I just found it useless for even basic chatbot-style back-and-forth. For example, I was combating some awful Chinese bot traffic on this website early in June and wanted to set up a Cloudflare filter to screen visitors. Siri AI pulled old, unhelpful documentation from some shoddy AI-generated webpage; ChatGPT told me exactly how to do it and what to do if it failed. Siri AI is an LLM-powered digital assistant down to the little interface details: the mini window, its brevity, the web references, and the integration with Spotlight. It would be an assistant first and foremost by design even if it used frontier-class models.

The fortunate news is that Siri AI is the best digital assistant available. The crème de la crème of the new architecture — and what propels Siri AI from a mediocre ChatGPT alternative of three years ago to one of the most elegant, transformative pieces of software — is the personal context, derived from Apple’s semantic index. Apple did not bury the lede about where the semantic index comes from: Siri and Spotlight are joined at the hip because the Spotlight index and the semantic index are the same. When Siri AI is first enabled on a new device, Spotlight takes up to a week to index files, notes, emails, and much more. The personal context exposes this information to Siri AI, including data from third-party apps that opt to donate to the semantic index. If Siri determines that it must use a particular note in Apple Notes, the system orchestrator will find the note in the personal context and provide it to the model for processing.

It’s self-evident why Spotlight indexing takes so long after updating to iOS 27: The system must index virtually every bit of information available across every app with a high degree of specificity. If it needs a text message from a decade ago to answer a question, the system should be able to find it in reasonable time. Siri AI falls apart without this functionality. And the model should be able to query the personal context with natural language — the system would be useless if it required precise search terms, hence the use of a semantic index rather than the lexical Spotlight index. If a certain Bloomberg reporter ever compiles The Siri Files from the past two years, I’d imagine the personal context would be known as the most complex part of the architecture. It is legitimately flawless in every test I’ve tried — iOS just seems to know me better than I do. In practice, Siri was able to find messages that even the Messages app’s search function couldn’t, notes whose titles I didn’t know, and poorly formatted emails from years ago.

This is only possible due to Siri AI’s LLM architecture. As I wrote earlier, the benefit of an LLM is not purely that it has a lossy recollection of the world’s knowledge, but that it has a grasp of human language. Siri AI captures intent better than any modern AI system — it was tastefully pre-trained to understand nuance and sometimes incomprehensible speech. “Find that note about note-taking apps in 2019-ish, maybe August.” This is a legitimate transcript of a Siri conversation I had, and the system found the note in mere seconds. It was actually from August 2020, and the note didn’t mention the phrase “note-taking apps” once. Another request: “Remind me to water the plants before my appointment tomorrow.” I had never put the appointment on my calendar — Siri found the appointment in my email after noticing it wasn’t in Calendar and set the reminder 30 minutes beforehand. The system is both persistent and clever, wonted qualities of a great assistant.

A perennial frustration that all iPhone users have likely had with Siri at least once is that talking to it felt like walking on eggshells. The new Siri AI is remarkably adept at understanding the messiest questions, whether it’s searching for a bit of niche information online, recalling the personal context, or calling an App Intent. There’s no longer a need to ask in a particular format: Requests like “What’s 4 p.m. England time?” are interpreted well. Siri AI understood I was asking for the conversion from 4 p.m. British Summer Time to Eastern time; the old Siri would downright give up (“I’m not sure I understand”) or just state the current time in England for such a query. Telling Siri, “I watered the plants,” instructs it to mark the “water the plants” reminder as completed. “I watered the shrubs” also works, even if the reminder only has “plants” in the title. If Siri AI ever needs clarification — such as if dictation couldn’t capture the full sentence — it proactively asks: “Just to confirm, did you mean X or Y?”

Siri AI has some interesting limitations on the Mac; namely, it doesn’t have full access to the file system and shell, unlike AI coding agents. I think this is a baffling limitation, especially because Siri AI could become the de facto tool for people — particularly power users likely to use the Mac for development work — to tweak settings on their computers. I use Claude Code all the time to modify property list files, organize and rename files on my desktop, or find things to delete. Just the other day, I asked Claude to find which “Developer Tools” were occupying 100 gigabytes on my disk — Claude found the deprecated simulators I had installed years ago and deleted them for me. I realize this is a niche request, but I think it’s important for an agent like Siri AI to interact with the file system, similar to Claude Cowork. I could see many people using such functionality to clean up their computers and organize files, especially if it were in the safe, trusted hands of Apple software. Siri AI is limited to finding files and cannot move or delete them. It also can’t create or edit files, nor can it use AppleScript, a legacy scripting language used to automate some apps. It’s a frustrating impediment.

Siri AI on the Mac.

Siri AI instead has an astute on-screen awareness capability. Out of curiosity, I opened a photo from the television show “The Office” — the one where Pam Beesly says, “They’re the same picture” — and asked Siri, “What episode is this from?” Siri responded in seconds: “The image features the character Pam Beesly (played by Jenna Fischer) in a scene from the Season 7 finale, ‘Search Committee.’” It even cited four separate internet sources, including a template of the infamous meme. The speed at which the model retrieved the data is due to the innate efficiency of an on-device model, but having it embedded within the system in this fashion is exceedingly helpful. On-screen awareness can analyze the full contents of what’s open — like a full article or note that trails off-screen — and goes beyond sending a simple screenshot to Siri. Siri AI also knows about currently playing media, even from third-party apps that don’t yet have App Intents.

Siri AI’s on-screen awareness feature.

When Siri must perform actions within apps — such as marking a reminder as complete, as opposed to listing unfinished reminders — it uses an App Intent. An App Intent is a small piece of code that, when run, can quietly launch an app in the background to perform a task. The intent works as a bridge between the app action (e.g., marking a reminder complete) and the system, which calls that action. App Intents are ubiquitous in iOS and macOS today: They’re required for Shortcuts and interactive widgets, like those in Control Center. Siri AI can now call these intents with better precision and semantic understanding — it can write complete notes with paragraphs and formatting or create usefully named reminders. When Siri uses an intent, a small “window” into the app’s content appears — such as a list of reminders or a new note — indicating a particular app was used. Broadly, Siri can now fully take action within apps on a user’s behalf, realizing — to a certain promising extent — Silicon Valley’s end goal of agentic experiences. The only catch is that every action in every app must be an intent to be usable by Siri.

Most native apps have rich support for App Intents in the beta. For example, I asked Siri to “convert this dashed list to bullets” with a note open in Apple Notes, and it obliged. Another time, Siri summarized an article I was reading in Safari and included a link and the author’s name in a note. Requests like these required a mix of on-screen awareness, personal context retrieval, and an App Intent to write to that note. I’ve generally found Siri’s ability to find relevant documents and match my formatting and writing style to be one of its best use cases, though the model occasionally outputs unformatted Markdown or misaligned paragraphs³. Siri also chooses, as best it can, which app’s intents to call without manually specifying, but currently defaults to using first-party apps. I must specify every time that Siri should add a reminder to Things rather than Reminders, for example. I hope these preferences eventually sync through the personal context.

Triggering App Intents and the personal context from Siri AI.

Developers contribute to the personal context by using App Schemas to mark certain information as indexable and add app actions through Intent Schemas; they can tell the system about open views and documents through NSUserActivity and App Entities. I think adopting these technologies will be table stakes as iOS 27 rolls out to the public. In 2024, I feared a lack of third-party App Intent support would hamstring the Apple Intelligence initiative after Apple Vision Pro suffered the same fate. But after using Siri AI, I really believe most users will eagerly await support for it in all of their favorite apps, including corporate developers like Google and Meta. Independent apps like my favorites will probably have support on Day 1. Uber, for instance, does indeed have an incentive to get people into the Uber app, but it has a strong record of using native iOS features, like Live Activities. The Uber app today has an integration with Siri to quickly book a ride, and ride-share information from the Uber app is listed in Apple Maps. These are voluntary integrations Uber has decided are good for its business, and rightly so. I think if Uber doesn’t support Siri AI within a few months of its launch, it’ll be at a disadvantage — I expect nothing less from the company that shipped a ChatGPT app almost immediately.

Convincing corporate developers to donate data to the personal context will be a tougher sell, though more plausible than it was in 2024. Meta’s apps, like Threads, WhatsApp, and Instagram, have already adopted Liquid Glass, though they undoubtedly use custom interfaces that are a hassle to rip out and redesign. People weren’t even particularly clamoring for Liquid Glass redesigns of these apps — the sentiment around the redesign is generally neutral to negative — but Meta spent time doing it. Apple still has a chokehold on app design and UI paradigms, and it would be Meta’s loss not to give the personal context access to Instagram direct messages or WhatsApp chats. If anything, wouldn’t that encourage more people to use third-party messaging apps, since they’d act indistinguishably from Messages? Companies like Meta and Spotify have lobbied governments for decades for the same access to system experiences as first-party apps — why wouldn’t they take advantage of it now that it exists?

Perhaps this is naïve optimism, but over my month with Siri AI, it legitimately felt indispensable. Through clever product direction, Apple has leveraged the candidly mediocre LLMs from the Google partnership to build one of the most elegant, user-friendly, and useful AI products from Silicon Valley in the last four years. I find myself gravitating toward Siri AI for all sorts of small tasks I was typing into Google or using ChatGPT for: quick facts, image lookup, and basic web searches. Local AI is Apple’s power play, and this makes Siri the most useful, delightful, and convenient AI assistant soon to be available on the market.

But the genuinely innovative personal context and App Intents might just be the panacea Apple needs not only to catch up in the consumer AI space, but to lead it. Apple has constructed a moat so powerful it has attracted the ire of governments around the world: the App Store and iOS. App Store customers demand high-quality apps that integrate with system functions. When Siri AI hits millions of devices in the fall, people will expect their favorite apps to support it, and if they don’t, people might be nudged toward first-party apps instead. I really do think Siri AI is that compelling — it erases decades of frustration people have had with their devices. Siri AI is Spotlight search for the natural-language era of computing. However badly OpenAI wants it to, ChatGPT can never build a moat as formidable as Apple’s. Apple’s bespoke advantage is not only its impeccable product direction, but its grip on the app developer market.

Like any generative artificial intelligence system, Siri AI makes mistakes, has limitations, and often performs worse than its competitors. I just can’t dismiss these as beta bugs — they’re deeply rooted in Siri AI’s models, architecture, and marketed capabilities. As I’ve argued many times in this section, Siri AI, for now, isn’t a replacement for the LLM chatbots people are used to, and it won’t be able to do complex work. It will never do complex work; Apple has never tried to wrangle the professional AI space. But even in its new niche — sitting alongside obscure products like OpenClaw, Manus, and Perplexity Computer — Siri AI is only months away from falling behind again, as the mainstream AI companies inevitably develop some answer to it at breakneck pace. So, if I could use only one word to describe Siri AI, I’d say, “promising.” It allays fears that Apple hasn’t been working hard enough, and it brings something useful and novel to the table. That’s a good place for Apple to be, but it’s still nascent.

I’ll be closely monitoring Siri AI’s development and App Intents adoption in the months and years ahead. I expect new model updates every few months — with or without Google’s help — and I expect Siri AI to progressively gain new capabilities over time. That’s the promise of generative AI systems: they get better and more useful through incremental model releases. If everything goes to plan, Apple has a winner on its hands. The “more personalized Siri” was worth the wait.

Siri’s new voice, ChatGPT extension, and app.

Apple Intelligence

There is significant ambiguity about what counts as Apple Intelligence and Siri AI as of this year’s operating systems. This isn’t an official answer, but I speculate that “Apple Intelligence” is Apple’s brand name for all of its generative AI products, like Visual Intelligence, Image Playground, and Genmoji, while Siri AI is just one of those products housed under Apple Intelligence. (It is unclear whether the “AI” in “Siri AI” stands for artificial intelligence, or Apple Intelligence.) The honest answer is probably that Apple itself doesn’t know, and that it hastily came up with the name “Apple Intelligence” in 2024 to tout the suite as a more private, helpful version of generative AI.

Regardless, the title of this section is certainly a misnomer, because Apple blurs the line between Siri AI and Apple Intelligence far too often in this year’s operating systems. There is a subtle fifth way to invoke Siri AI on all platforms: context menus. When you select any content in iOS, macOS, or visionOS, an Ask Siri button appears at the beginning of the context menu. Tapping Ask Siri is supposed to attach files or text to a new Siri AI chat, where you can ask questions about the content. As of Beta 3, this functionality is somewhat broken: Using Ask Siri usually does not attach files, and when it does, it sometimes attaches the wrong file. It works reliably only when text is selected in first-party apps, or in third-party ones that use native text fields. This disqualifies every code editor on the market, including Xcode. There are two possible explanations for these rough edges: (a) the system suffers from simple logic errors and will be remedied before public release, or (b) the Ask Siri field is in more places than it should be, and that will be addressed in a future beta. It’s probably some combination of the two, since I think text, images, and files in Finder should be attachable, while apps on the Home Screen of iOS, for example, shouldn’t be. This is a silly but real example.

The Ask Siri menu item on iOS.

I’m including this here only because the incessant, almost Copilot-like Ask Siri field replaces Writing Tools, one of the most disappointing features from the Apple Intelligence release two years ago. There is no longer a defined set of functions Apple Intelligence can perform in text fields, such as Rewrite, Proofread, and Summarize; instead, you must ask Siri for them. Apple’s models now handle all requests; the ChatGPT integration has been removed. In a supported text field, just open the Ask Siri field or tap Write with Siri in the QuickType bar on iOS⁴ and prompt for additions or changes. Siri will then insert the desired changes into the document. It can also summarize content or, cleverly, insert a summary into a new note — all Siri AI functionality carries over. But this system does not address my key frustration with the previous Writing Tools implementation: It neither shows changes nor lets users approve or reject them. For instance, if I ask Siri to make a change in a long document, it will not indicate where it made the change or ask me to approve it. I would have to hunt the change down manually. Vexing.

Write With Siri on iOS.

Siri AI, in the current beta, immediately inserts changes into the document, which is both a regression and an improvement. In most apps, Writing Tools would display the entire edited text in a pop-up, which was next to useless. However, I do think the new functionality is presumptuous — there should be a way to approve or reject changes before Siri inserts them into the document. I am glad that there is consistency across apps: The experience is the same regardless of whether you’re working in a first- or third-party app. Previously, Writing Tools would use an error-prone, diff-like interface to review changes only in some first-party apps, TextEdit being the only one where I could reliably trigger it. Other apps would use the pop-up, and neither was a particularly useful implementation.

Write With Siri on macOS.

The replacement for “Proofread” is, mercifully, much more elegant, though still constrained. The new feature replaces the system grammar checker on both iOS and macOS, though I forgive those who don’t remember there was already one built into the operating systems. The grammar checker has always worked automatically alongside spellcheck and, on iOS, is integrated with the autocorrection system. The new feature works much the same way on macOS, underlining grammatically incorrect phrases as you type and displaying suggested replacements in the Control-click context menu. On iOS, there’s now a screen that lists all suggestions the system has made, with options to accept or ignore individual ones. This addresses my biggest gripe with Writing Tools, as explained earlier. To access it, select a suggestion as on macOS, then tap the magnifying glass icon in the Select menu. I haven’t a clue why this extremely convenient view — which even displays the reason for the correction (“incorrect apostrophe”) — is unavailable on macOS.

The new feature is slower than the previous algorithm. Corrections usually take a few seconds to about half a minute to appear, whereas the previous grammar checker was almost instant. Sometimes, the system is unusually sluggish and requires a manual trigger in Edit → Spelling and Grammar → Check Document Now. (It isn’t currently possible to trigger it manually on iOS.) Anecdotally, I don’t find the suggestions from the new system to be significantly more advanced than those of the prior checker. It still mostly corrects verb agreement, tense, and repeated words, but pales in comparison to even the free version of Grammarly, which I begrudgingly run on my Mac to catch embarrassing mistakes. Grammarly is also much faster, since it uses a complex algorithm rather than an LLM for most suggestions. For more precise editing — where intent and writing style must be intricately understood — the new system doesn’t even try, and Siri AI’s models are not large enough to act as a standalone editor; LLM chatbots are better suited for such edits. Regretfully, I cannot recommend this new system over Grammarly — it is too limited and slow to be useful.

Proofread on iOS.

These three features together — attaching files and text as context, asking Siri to edit or generate new content, and the new grammar checker — comprise the terribly named Write with Siri suite. Write with Siri is easily the most unintelligible group of features introduced in this year’s operating systems, and it deserves a lot of improvement. To briefly summarize:

There is a new Ask Siri field in every context menu in all operating systems, sans watchOS, of course. Selecting text, an image, or a file should attach it to a new Siri chat, where you can ask questions about the content. All other Siri features, like App Intents and the personal context, remain available in this mode.
Siri can also edit text if it is opened in a supported text field — just open the Ask Siri field by selecting text to attach as context, or open the field from a new line. If no content is selected, Siri can still edit the document and add new content. On iOS, iPadOS, and visionOS, a new Write with Siri button sits pinned to the QuickType bar and emulates this functionality.
The “Proofread” feature from Writing Tools is now built into the system grammar checker — suggestions are made automatically. Only on iOS, iPadOS, and visionOS is there a diff-style interface to view all suggestions in a document and approve or reject them: select a suggestion and tap the magnifying glass icon. This interface is not available on macOS.

If this is the year of polish for Apple’s operating systems, Write with Siri leaves a lot to be desired. I do think this year’s features are better conceived than Writing Tools, but that’s such a low bar to clear. It makes sense to integrate them systemwide, but the result is a patchwork of ways to prompt Siri. That itself isn’t bad, but I think the offerings are more convoluted this year, and that most people will either be frustrated by the Ask Siri prompts littered throughout the system or simply confused when the operating systems ship in the fall.

Visual Intelligence, unlike Writing Tools, has not been completely axed from the operating systems — but it now does business as Siri AI. Previously, Visual Intelligence was available in the screenshot menu and as a camera mode invoked by pressing and holding Camera Control on iPhone 16 and later. The former, introduced last year, still exists on iOS with a nearly identical interface, presenting two buttons: Ask and Search. Search is unchanged and performs a reverse image search on the screenshot; Ask, which previously used ChatGPT, now attaches the screenshot to a new Siri AI chat. This is akin to the infamous Ask Siri text field covered earlier in this section. I seldom used this feature in iOS 26 because it was slow and the ChatGPT integration was anemic, and I still don’t use it because it’s only a more roundabout way to trigger Siri’s on-screen awareness. I instead find myself opening Siri AI directly from the Dynamic Island and asking about what’s on-screen, rather than using Visual Intelligence. But I understand the Ask button is there for consistency.

Visual Intelligence through the screenshot menu on iOS.

Siri also brings Visual Intelligence to the Mac, but, just like on iOS, Apple brands it as Siri in the Screenshot app. Screenshot is an app on macOS, typically invoked using the keyboard shortcut Shift-Command-5. There is a new Ask Siri option in the screenshot menu, which takes a screenshot of a selected area and attaches it to Siri automatically; it can also be opened by pressing Shift-Command-6. If you use the keyboard shortcut — which highlights the selected area in an eye-watering, high-dynamic-range rainbow glow — there is also an Image Search button, again like iOS. I use CleanShot X for my screenshots, so I haven’t used this feature much, but I also haven’t missed it, since Siri AI on the Mac has the same automatic on-screen awareness as on iOS. I’d rather trigger Siri AI and prompt it than open Screenshot to do the same thing.

Visual Intelligence on macOS.

The camera mode also eschews the Visual Intelligence brand name. I think it’s disingenuous to call these Visual Intelligence features, but that’s what Apple calls them on its website. My working theory is that “Visual Intelligence” refers to Siri AI’s vision capabilities, but I wouldn’t be surprised if Apple phases out this branding entirely next year. It makes no practical difference for people, since the “Visual Intelligence” label is not used once in the operating systems, only in marketing. Siri is now a literal mode in the Camera app, listed alongside Photo, Video, Spatial, etc., and can still be opened by pressing and holding Camera Control. The familiar Ask and Search buttons flank the shutter button. Snapping a photo immediately sends it to Siri, whereas the Ask button takes a photo, then allows additional prompting before sending the image in a new Siri chat, similar to the screenshot menu. Images sent via Visual Intelligence are always saved to the Siri app for later viewing, and you can ask follow-up questions within the Camera app. This mode is, amusingly, also available on Apple Vision Pro, and I’ll admit it feels like something out of a science fiction movie when I use it. I think this mode is — at least on iOS — the best way to attach images to Siri AI, but I find the models worse at image processing than traditional chatbots.

The Siri mode also introduces what Apple calls “smart actions”: enhanced previews, for example, to view the nutritional information of a dish or add a custom pass to the Wallet app. My only gripe with these is that I think there’s a lot of untapped potential here for personal software. The feature is limited to only certain actions, but when they work, they work almost uncannily well — Siri could detect what food was in an image with astonishing accuracy because it’s LLM-powered. But I don’t see why these must be limited to Apple-provided use cases. I’d love, for instance, the ability to see how much sunshine and water a plant needs by just snapping a photo of it. Perhaps a price-checker would be handy, with options to view the price history of an item on select sites. People are writing (“vibe-coding”) apps like these with generative AI all the time, so an interesting way to leverage these models as they get smarter would be to have them create custom smart actions on their own. Smart actions themselves are just pleasant user interfaces around a response the model provides — I think Apple could have some fun with them.

The Siri mode in the Camera app, and smart actions.

Apple is already dabbling with personal software in these operating systems. There are two such features, both fascinating: Describe a Shortcut and Describe Extension.

Describe a Shortcut works, naturally, in the Shortcuts app. There’s a new Apple Intelligence tab in the Shortcuts editor with a simple text box. The editor now defaults to this tab instead of launching into actions, but this can be changed in Settings. Tell Apple Intelligence what the shortcut should do, and it will churn for a few seconds and return with a hopefully helpful automation. The feature is well suited for simple shortcuts, which really should be the primary objective — things like sending text messages or setting reminders, which most people would probably like to automate. Shortcuts has always been one of those oddball Apple apps that’s so helpful yet so daunting to novices, and Describe a Shortcut eliminates the barrier to entry. It does fall apart with more complex actions like AppleScript or loops — the model is trained on Shortcuts actions, not code, and probably wouldn’t be able to build something like Federico Viticci’s famous Apple Frames shortcut, which uses hundreds of actions strung together to frame screenshots. Viticci, the editor in chief of MacStories, instead made a command line utility that uses better models and can generate more complex shortcuts, and I think that’s better suited for professional Shortcuts work.

Describe a Shortcut also summarizes existing shortcuts.

But the largest barrier of all is computer code, such as JavaScript, HTML, CSS, and Swift — historically the only ways to write Safari extensions to customize websites. Describe Extension is Apple’s first foray into agentic programming with its own models, and I think the feature is transfixing in action. At its core, Describe Extension can modify websites however you’d like through natural language. Prompts like “Remove the search bar from this page” work perfectly. I’ve always injected my own handmade CSS into certain sites — infamously X — through Jeff Johnson’s excellent extension StopTheMadness to remove specific elements, but Describe Extension obviates the need for such code. Custom extensions sync automatically across devices, too. The agent can also create new content on a page, like a timer or custom UI. When you write a prompt, the system will, peculiarly, recommend existing extensions on the App Store that may already work. If you continue, Apple Intelligence will take a few seconds to generate the extension. You can then try it out and save it, edit it, or discard it. The code the model writes is not viewable.⁵

This feature is so delightful. It is virtually limitless: I asked it to “generate an extension where, when activated, it copies selected text as Markdown and encloses it in a block quotation.” I do this all the time for posts on this website. The model churned for a few seconds and came back with a nearly functioning extension: When I selected some text on a webpage and clicked the icon in the toolbar, it did almost exactly what it was supposed to, with the slight hitch that it didn’t handle new line characters correctly. I then told it about the oversight, and a few seconds later, the extension worked like a charm. Much like Siri AI itself, this feature is just so whimsical and a joy to use. It even chose an appropriate clipboard icon for the extension by itself, so it doesn’t look out of place in the Safari toolbar. You can confine an extension to a single site by asking the agent — otherwise, they’ll work on all sites. It’s aware of when the extension should only apply to a particular site, such as if asked for a specific enough change. Customizing websites is now only a few keystrokes away in Safari.

Describing an extension on iOS.

The Apple Intelligence suite this year is fortuitously much stronger than two years ago. But I still feel that most of the features were a bit rushed — I get the sense that Apple wanted to integrate Siri throughout the system but aimed too ambitiously. Ask Siri prompts litter every context menu and software keyboard in the operating systems. The new Proofread feature had real potential to compete with Grammarly — my archnemesis, and that of many other writers — but it isn’t even consistent across platforms. The Siri mode in the Camera app is clever but limited in ways that don’t make sense in the age of personal software, which Apple astutely and ingeniously embraced in Shortcuts and Safari. Don’t mistake this criticism for apathy, but I think the overall Apple Intelligence suite is still underwhelming and behind where it should be. It’s gratifying and innovative where I wasn’t expecting it to be — like Describe Extension — but underwhelms where it shouldn’t. It’s a mixed bag for sure, and it’s a real head-scratcher that Apple didn’t invest more time into getting it right, as it did with Siri AI.

Design

As I wrote at the beginning of this article, the first real feature announcement at this year’s keynote was a slider to adjust the opacity of Liquid Glass. The slider marked the beginning of what I’d call Apple’s Liquid Glass apology tour, with a litany of small tweaks and refinements to last year’s bold redesign. Liquid Glass was interesting and controversial from the start: I liked the material and the overall refresh of common controls, but I eventually admitted late in the beta process that the redesign had been too ambitious and buggy for final release. Tab bars in iOS 26 were simply illegible; toolbars blended in too much with content underneath; sidebars in macOS 26 Tahoe were atrocious; and the operating systems reeked of incompetence, with tacky icons accompanying menu items and inconsistent window corner radii. It appeared the designers of the 26 operating systems had spent too much time musing over their glass trinkets to care about the user experience.

After the perfunctory December departure of Alan Dye, Apple’s software design chief responsible for Liquid Glass, it appears saner heads have prevailed at Apple Park once again. The new operating systems look as if Apple put care, thought, and time into them. They’re not a return to the pre-OS X 10.9 Mavericks heyday of software design, but they get closer than I predicted. I’m pleased that Apple didn’t just timidly dial back the rotten parts of the 2025 redesign — it scuttled the designs that weren’t working and returned to form in many interesting ways, so much so that I had to dedicate a section of this article to documenting what has changed. The core thesis — both mine and presumably Apple’s — is that reliability was deservedly the focus in every one of this year’s design updates.

On all platforms, Apple has now modified the eponymous material used throughout the redesign — from toolbars and tab bars to prominent buttons — to increase legibility. The default Liquid Glass appearance is more opaque, frosted, and tinted. The prior material emphasized showing content underneath — like a pane of clear glass — and relied on a diffraction layer to blur it enough to retain legibility. This was a dreadful design: While attractive, it treated toolbars and tab bars as second-class, ignoring that whatever was underneath was irrelevant to a person interacting with a toolbar. It showed too much extraneous content. The new Liquid Glass material darkens content that shines through and adds a new soft blur layer. The best way I can describe it is that the old material was like a clear window, while the new one is more like privacy glass — the blur and darkness conceal the content underneath. The diffraction layer, therefore, plays a smaller role in maintaining clarity, leading to a much more pragmatic appearance. In the keynote, Apple said Liquid Glass now “diffuses complex content behind it much more effectively.”

The new default Liquid Glass appearance across apps.

The Liquid Glass slider is a genuinely peculiar feature. By default, the slider is set to the middle position. When you slide left from center, the material gets less blurry, with the far-left selection almost eliminating the soft blur layer. When you slide right from center, the blur layer doesn’t change; the material becomes more solid and eventually becomes opaque. The left half of the slider adjusts the blur layer, and the right half adjusts the opacity. This doesn’t matter to almost anyone except the most astute UI designers and nerds, like the ones reading this article — left means clear and right means opaque — but it gleans some insight into the four distinct layers of Liquid Glass: glass, blur, tint, and diffraction. Anecdotally, I think most people in the fall will, if given the option during onboarding, choose either the default or somewhere on the right side of the spectrum, since the changes are most noticeable there. The material really does get much darker, almost reminiscent of iOS 18 and earlier. I’ve tested many different settings and prefer mine in the middle of the left section.

The Liquid Glass slider in light and dark mode.

Liquid Glass elements also feature a new border and drop shadow that distinguish them from other content. These layers hug the material around the edges, giving it an almost 3D appearance. The first time I saw the new material after updating my iPhone to the iOS 27 beta, I was struck by how similar it looked to Aqua, the design language introduced in the first version of Mac OS X by Steve Jobs, who famously noted that “one of the design goals was when you saw it you wanted to lick it.” Aqua was distinctive in how UI elements cast shadows on their backgrounds, like real drops of water. The refined Liquid Glass material in the 27 releases applies this same shadow and depth effect, and it’s axiomatic where Apple’s design team — perhaps with more liberty to look back to the Jobs era — drew inspiration. The resemblance between the early-aughts-era Aqua and the 2020s-era Liquid Glass is almost uncanny in these releases. It looks gorgeous, legible, and practical, even after dialing down the opacity and blur. I’d argue it actually looks better with minimal opacity — something I’d never thought I could say about Liquid Glass, all because of the new drop shadow and improved diffraction layer.

One of the foremost gripes with the Liquid Glass redesign was the new toolbar design — plainly, that there was no longer a distinct toolbar. This “design” was abominable on iOS because it eliminated the border between content and buttons, but it was truly accursed on the Mac. That change has been reversed in all of this year’s operating systems: The toolbar background is again nearly opaque, using the same material it always has in iOS, iPadOS, and macOS. Liquid Glass buttons sit atop the toolbar and show almost no background content — even when the slider is pushed to its leftmost position — drawing a clear line between the toolbar and the content below. Tab bars still use the Liquid Glass material, but I find their appearance far less objectionable because they’re encapsulated within Liquid Glass — the material itself distinguishes the tab bar, even more so in the 27 operating systems. The old toolbar merely floated atop the content with little to no visual separation, only a light blur. It destroyed the interface hierarchy between views and controls. The new toolbar leaves no ambiguity about whether the content at the top of the screen is part of the view itself or the toolbar.

Toolbars in iOS 27.

Toolbars in macOS Golden Gate.

The new-old toolbar design has drawn some ire online from people who seemingly don’t have a clue about user interface design; they claim it doesn’t honor the transparent nature of Liquid Glass. But that was the biggest problem with Liquid Glass: It treated everything, including elements that ought to be elevated above views, as hierarchically equal. Candidly, the people who first designed it were either uninterested in the ramifications of blasting a UI with semitransparent materials, or failed to test their design properly in all apps and use cases. I must assume it’s some amalgam of the two.

On macOS and iPadOS, specific design changes are even more noticeable and appreciated. One of my callouts in last year’s OS review and my “Can Apple Fix This in 6 Weeks?” postscript criticizing the redesign’s shortcomings was the sidebar design in macOS Tahoe and iPadOS 26. Sidebars “floated” above content, and Apple’s own Human Interface Guidelines encouraged developers to extend their content underneath the sidebar. Horizontal list views would push their content past the sidebar to the edge of the window, and the HIG said to scale up media knowing that the sidebar would partially obscure it. The sidebar design was a complete disaster, and may have been one of the worst designs Apple’s typically star-studded Human Interface team has ever cooked up. This design is now gone in macOS Golden Gate and iPadOS 27, replaced by a near-replica of the macOS 11 Big Sur-era sidebar. There is again a distinct border between the sidebar and the detail content, and the sidebar is no longer inset. It’s instantly recognizable as soon as you boot into macOS Golden Gate — it just looks and feels right.

The Music app’s sidebars in macOS Golden Gate.

All apps — including those that aren’t compiled for the new operating systems — receive these new design improvements. Third-party apps in macOS Golden Gate look great with no developer work, especially the ones that put significant care into their macOS Tahoe redesigns. The same goes for iOS and iPadOS, though some developers put extra work into making their apps look more habitable on iOS 26, and now that Apple has addressed many of the design shortcomings, those considerations are obsolete. I’d say most developers will have to put little to no work into finalizing their apps for the new releases in the fall. All apps on macOS — including those that haven’t even been redesigned for macOS Tahoe — use a new, tighter corner radius that truly makes me happy. It’s slightly more aggressive than macOS 15 Sequoia, the pre-Liquid Glass version, but a striking improvement over the bulbous, Fisher-Price-esque windows of macOS Tahoe. macOS Golden Gate just looks like it was designed by adults who know what they’re doing. Apple’s vapid obsession with “concentricity” is gone, and the design looks thoughtfully crafted.

The menu item icons are mostly gone in macOS Golden Gate.

The unified window corner radii in macOS Golden Gate.

Every aspect of the design has been given a once-over with a fine-tooth comb. The incessant menu item icons are now gone by default. Notifications now adapt better to light-colored Lock Screen wallpapers and even drop down from the top-left corner to indicate the new Notification Center gesture. Animations throughout the system are smooth as butter — I have yet to notice even a frame drop while using the first few betas. Sidebar icons on iPadOS and macOS are tinted again and even fade to gray when the app isn’t in the foreground, just like they have for decades. App icons created through Icon Composer are automatically rendered to remove the “fuzziness” that plagued them last year, with no developer intervention necessary⁶. I could go on about all of the little details that make the new operating systems feel like they’ve been crafted with care.

Liquid Glass in macOS Golden Gate, using my preferred clearer appearance.

I can’t overstate how many people have come up to me — strangers and acquaintances — in the last year to complain outright about the iOS 26 design. The criticism is harsh but not undeserved: iOS 26 by default is almost unreadable, and macOS Tahoe truly is one of the worst macOS releases in a long while. People couldn’t even read the time on the Lock Screen without switching to the Solid clock appearance, for heaven’s sake. The updated Liquid Glass design in iOS 27, iPadOS 27, and macOS Golden Gate goes a long way toward addressing the havoc of last year’s rushed releases and cherishes the fresh coat of paint I and other UI design enthusiasts praised in our reviews of the new UI paradigm. I think this will be the year most corporate third-party developers will adopt the new design, especially since they’ll be required to compile their apps with Liquid Glass beginning in the fall. It’s allaying to see Apple take developer and user feedback to heart.

Photo Editing

I have unwavering respect for my readers, so this section does not contain a fourth cathartic tangent deriding AI-generated imagery. But Apple has added some new generative AI image-editing tools across the system that are worth briefly covering.

Two years ago, Apple announced Clean Up in the Photos app, a generative AI feature for removing unwanted objects from images. This year, the Photos app improves Clean Up with new models and adds two AI photo-editing tools in an apparent effort to better compete with Google’s offerings: Extend and Reframe. The new modes are housed in the new Tools menu of the photo editor on all platforms, and all work with varying levels of success.

The enhanced Clean Up tool uses two new models: Fast and High Quality. Users can choose between the two manually, or select Auto, which recommends the best model depending on the task. The Fast model works as soon as an object is highlighted in the image, editing the photo in only a few seconds once the model loads. Its accuracy, I’ve found, is no better than that of the prior Clean Up model, and it still produces sloppy images; I wouldn’t recommend using it for much of anything, even small touch-ups. The larger, High Quality model appears to send images to the cloud and takes much longer — up to a few minutes — to make edits. When it’s selected, it asks you to mark all edits at once, then tap the Clean Up button when finished to let the model go to work, so you aren’t stuck waiting separately for each edit. The results are uncannily good — the larger model is worth the wait, especially with the batch-processing interface.

The improved Clean Up feature in iOS 27.

The Extend and Reframe features are much less useful. Extend expands the photo around all four edges, using what’s already in the photo to craft a realistic scene. When you first enter Extend, it initially displays the original photo with blurred, “dreamy” edges. You can then expand the photo using the drag handles, or by pinching to zoom out. I’ve found that the feature can generate a new image up to double the size of the initial one, but it isn’t possible to limit Extend to only one side of the image. Users cannot, for instance, strictly extend the top edge without altering the other three sides to some degree⁷. After tapping the Extend button, it takes a few minutes to generate the new photo. It’s far riskier than Clean Up, especially because it works with every image, including highly complex ones where it would be difficult, if not impossible, to predict what was outside the frame. Extend will, for instance, give pants to someone wearing shorts or create people who don’t exist in the background. Every photo modified with it clearly looks AI-generated — nobody will be fooled. Extend also replaces the Extend Background feature on the Lock Screen for ill-fitting photos, and I think that’s perhaps the only place it’s suitable.

The Extend feature in iOS 27.

Reframe works much the same way, but I think it has even less utility. It’s effectively the same feature as Extend — presumably using the same models in the cloud — but it also alters the Z-axis, or depth. It first takes a few seconds to create a depth map of the image, then lets you pan around using the same “blurry edges” interface from Extend. After a few minutes, it “fills in the gaps,” as Apple says, generating the missing edges to “reframe” the photo. It, too, is unconvincing to even a casual viewer. Images clearly look manipulated at best, or completely fake when the model struggles, and the degree of movement is quite limited. The models are nowhere near as capable as ChatGPT or Google’s Nano Banana image generator, built into Gemini. And I don’t think it’s anywhere near as useful as some of Google’s or Samsung’s AI image gimmicks, which, while sloppy, at least have some real-world value. Reframe and Extend don’t seem to me like very intentional or well-crafted features, and I don’t think they should’ve shipped in their current state, if at all.

The Reframe feature in iOS 27.

Image Playground, Apple’s app for creating AI-generated images, is much the same, though it adds a “photorealistic” style that really does look terrible. Perhaps it’s due to Image Playground’s limited availability, but I haven’t heard of a single person using it in the real world, whereas I do know image generators are rife among office workers for presentations and fliers. Free tools like Nano Banana and ChatGPT work better, so I keep asking myself why Apple intends to be in this market. Do Apple employees use it often internally? That doesn’t seem aligned with the company’s culture at all, so maybe a handful of executives who initially came up with the idea two years ago are intransigent about letting it go. I can’t see the new model being any more popular than the last one, given that even when Image Playground integrated ChatGPT last year, it didn’t seem to gain any more users. Image Playground ought to be shown the door.

Feelings aside, I just don’t think Apple is suited to deliver generative AI image features. Image generation is complicated — even for the major AI labs spending hundreds of billions of dollars on pre-training — and Apple’s models are not powerful enough to ship on the most popular smartphone in America. People will stumble upon these features while they’re editing photos on their new iPhones in the fall, but they’ll either find them useless or — worryingly — use them anyway, subjecting everyone to a flood of deceptive, poorly made content online. Apple felt internal and external pressure to ship image generation in 2024, but I don’t believe that pressure exists anymore. Clean Up is the sole exception to this: It was fine in 2024, and it’s good enough now. I don’t necessarily have a problem with people using it. But Extend, Reframe, and Image Playground don’t meet Apple’s standards, and their premise should be rethought entirely.

Miscellany

I began this review over 11,000 words ago with an intrepid claim: that this year’s operating systems hearkened back to the age of reliable yet innovative Apple software. Indeed, iOS 27 and macOS Golden Gate are the most reliable Apple platforms since iOS 12 and macOS 10.14 Mojave, when the company placed great emphasis on speed, performance, and bug fixes. Apple displayed a slide with close to 100 optimizations made this year, including a new Spotlight index, faster app launch times on older devices, and improved connectivity switching on iOS. I can’t remember the last time — if ever — the company spent so long describing changes to the CPU scheduler.

Everything in these releases feels faster and less buggy, even on my iPhone 17 Pro and relatively new M3 Max MacBook Pro from 2023. Everyone will feel the improvements, not just those using older devices. I’d argue they make even more of a difference than Siri AI and most of the other features introduced in the past few years. Some changes I’ve noticed include:

Faster and more reliable search in Mail, a feature that has languished for decades.
Correct results when searching System Settings, which has been partially broken since iOS 16 and macOS 13 Ventura.
Improved AirPods automatic device switching, which was too eager to switch to a Mac when AirPods were connected to an iPhone.
Faster and more reliable iPhone Mirroring.
Much better battery life — up to two hours more on iOS per charge and an hour more on my degraded MacBook Pro’s battery.
Much smoother performance while using Low Power Mode; it was almost unusable in iOS 26.
More honest Screen Time metrics on iOS and macOS, and improved syncing across devices.
More reliable automatic switching to cellular data when a Wi-Fi network is down; there is also a new Connectivity Assist, replacing Wi-Fi Assist, which worked sporadically.
System apps like Passwords now remember the sidebar size across launches on macOS.

There are some additional features I didn’t have space to cover elsewhere in this review.

The Passwords app will use an agentic computer use model to change compromised passwords on a user’s behalf. I don’t have any weak or compromised passwords to change — it’s only available if you do under Passwords → Security — but I am eager to test this out. This is one of Apple’s most ambitious generative AI features.
Keyboard dictation on iOS can now use the new, reliable dictation model that debuted last year in Notes and Voice Memos. I hoped for this in last year’s review, and it works like a charm. It’s only available for iPhone 17 and later models and iPads with M-series processors. Apple’s website indicates it should work on Macs with M3 or later chips, too, but it isn’t yet displayed under System Settings → Keyboard → Dictation → Advanced Dictation Preview, as of macOS Golden Gate Developer Beta 3.
Apple Intelligence now analyzes and summarizes HomeKit Secure Video footage in the Home app, and a new search bar finds relevant clips with natural language. I’ve found this feature hit-or-miss. Clips have a new appearance in the camera view, and there is a new unified All Cameras page with a list of clips from all video cameras in a home and a summary next to each clip. Perhaps it’s because my cameras are not 4K, but I’ve found the summaries unhelpful. Most are “Motion was detected,” “Someone was detected,” or, blankly, “A summary is not available for this video.” It seldom detects when packages were dropped off or if animals were spotted — I’ve never had this problem in previous versions.
The ChatGPT extension in Siri, released in 2024, hasn’t been removed. If enabled under Settings → Siri → ChatGPT, you can choose to send a prompt to ChatGPT in both the Siri app and Spotlight. (Voice queries are not routed to ChatGPT, and you must select ChatGPT every time; you cannot set it as the default assistant.) ChatGPT does not have access to the personal context and cannot call App Intents, which makes the Siri AI system unavailable in the European Union.
Messages and Mail now offer smart suggestions powered by Apple Intelligence, based on the message content. These suggestions allow people to add, say, a reminder in Reminders or a note in Apple Notes about the message when relevant. They work reliably, though I think it’s a shame that third-party apps can’t offer to be suggested apps or show suggestions themselves.
The Settings pane for connected AirPods has been redesigned. It also now includes a new equalizer, or EQ, selector to adjust the volume of mids, highs, and lows. The selector doesn’t significantly alter the sound profile, since AirPods already sound fantastic, but they tend to be mid-heavy, which doesn’t favor genres like hip-hop or rhythm and blues. I can see this being popular in the fall.
Nearly five years after the notch was first added to the MacBook Pro, macOS Golden Gate hides menu bar applets behind a double-chevron icon when the notch would otherwise obscure them. Without a menu bar manager app like Bartender or Vanilla, menu bar items that didn’t fit in the limited space to the right of the notch would disappear. This worsened when an app’s menu items overflowed to the right, further limiting the space for applets. If either of these scenarios occurs in macOS Golden Gate, macOS will instead automatically hide the overflowing items behind a new icon. (This temporarily breaks menu bar managers during the beta.) I cannot believe such a feature took so long — do Apple employees not use menu bar apps?

The menu bar chevron in macOS Golden Gate. (The blank space is the notch.)

macOS now keeps a more comprehensive list of apps that run in the background under System Settings → General → Login Items & Extensions. This list was previously limited to daemons in the ~/Library/LaunchAgents and ~/Library/LaunchDaemons folders; it now lists apps registered under the background task management database introduced in macOS Ventura⁸. From this list, it is possible only to disable these apps from running in the background. To remove them, such as after uninstalling the app, run sudo sfltool dumpbtm in Terminal, find the file paths of the apps you wish to remove, and rm them.

The background task management database in System Settings.

Apple Intelligence in Safari can automatically create special tab groups called topics. It works well enough.
Genmoji, the feature that allows people to create custom emoji stickers, has been updated with a new image generation model. The results are almost identical to those of previous versions, but the model is much easier to prompt. It adheres to instructions more closely, uses fewer words, and better understands intent. (I still don’t know of a single third-party app that supports Genmoji, despite the billboards plastered throughout Manhattan.) Genmoji is also fully available on macOS Golden Gate through the Messages app.
Image Playground can now generate wallpapers and is presented as an option alongside Photos, People, Color, etc., in the wallpaper chooser. It works just as well as anyone would expect, which is to say not well.

Image Playground wallpapers, new AirPods settings, and new Mail search.

This year’s review came in at nearly 4,900 words shorter than last year’s. For that reason, I don’t think Apple is likely to replicate this year’s releases for a while, probably half a decade at minimum. WWDC is a press event designed to garner attention — the company wants people and the press to get excited about how their existing iPhones will become even more capable and feature-rich in just a few months, all for free. And most years, that strategy does pay off. The contingent that resents iOS updates has slowly dwindled since iOS 7; almost everyone updates to the new version of iOS by spring. But I think Apple knows that iOS 26, while it had nearly identical adoption numbers on paper, was quietly resented. It did too much at the expense of usability.

I said this at the beginning, almost 13,000 words ago, and I’ll repeat myself: iOS 27 is a return to form for a company that just a few months ago felt like it had lost its way. Marred by politics — internal and external — regulatory restrictions, bad press, C-suite changes, and a heavily criticized two-year delay of generative AI features that were already late when they were announced, Apple was not in the best spot earlier this year. I think iOS and iPadOS 27, macOS Golden Gate, and Siri AI put the company in a much better position, building trust with the press, developers, and probably users. People will find Siri AI genuinely useful in all sorts of scenarios, from searching within apps to automating odd tasks. Each of the new platforms was designed with care and respect for the user experience, qualities unseen from Apple’s Human Interface team for far too long. There are missteps, but overall, this year is a great one for Apple software. I don’t say that every year, but I wish I did.

iOS 27, iPadOS 27, macOS Golden Gate, and Siri AI are available in public beta beginning Monday.

Siri’s new “expressive” voice is only available on certain devices. My preliminary impressions are that it’s not a significant improvement over the old voice, and it remains miles behind ChatGPT’s voice mode or Gemini Live. I’m not perturbed, though, since I never turn on the Speak Back for Voice Input option. ↩︎
Anecdotally, I find Siri AI’s hallucination rate is lower than that of GPT-5.5 Instant, OpenAI’s default model, though this could be because I haven’t been asking it tough questions. Siri AI does a fantastic job of providing links to external sources and often chooses to quote them rather than summarize or rewrite — an interesting and manifestly intentional design choice. As of Developer Beta 3, Siri AI also writes a disclaimer at the bottom of each response: “By the way, always verify important details. As an AI, I may make mistakes.” The disclaimer is written in plain text as part of the response, but it’s always reliably there. I wonder if Apple will move it into the interface instead, as some other chatbots have, or if it will conspicuously remain in the response. ↩︎
One of the only new features in Apple Notes this year is that it’s now possible to paste Markdown into Notes and have it automatically converted to rich text. My hunch is that Apple added this primarily for Siri AI, since it would’ve been harder and less extensible to train the model to write rich text over Markdown, the de facto language of every AI model. ↩︎
This must be a beta bug — it genuinely feels like something Microsoft would do. (Cupertino, start your photocopiers.) Simply put, you can either select text to provide it to Siri as context, or select no text at all and still invoke Siri. This works across platforms. The button in the QuickType bar, limited only to iOS, iPadOS, and visionOS, is redundant and does the same thing as selecting no text. ↩︎
It is viewable during the developer beta period, but you must submit a bug report to read the code. Feedback Assistant attaches it to the report automatically. ↩︎
App icons comprise three layers beginning in the 26 operating systems: the background, the glyphs, and Liquid Glass. Developers provide these elements individually to the system, which renders them on-device. The previous rendering algorithm didn’t properly distinguish the glyph from the background, making the glyph’s edges appear soft and blurry. In the new operating systems, edges look crisp and polished. The Liquid Glass effect on top has also been made less aggressive, and there’s an optional clear Liquid Glass material within icons, used by the Maps app. ↩︎
The blurry edges in Extend mode subtly indicate that it will always modify all four edges; I think Apple did a phenomenal job designing the UI. It is possible to request that there be more content on one side, but all four sides will be modified to some degree anyway. It’s slightly unintuitive, but I think the company did its best given its models’ capabilities. ↩︎
Apple’s support documentation disputes that BTM-registered apps weren’t shown in this list, but I think I’m correct. I indeed see many more apps in this list after updating to macOS Golden Gate, none of which are listed in the aforementioned Library subdirectories. They’re registered in the BTM database. My guess is that enterprise-configured Macs — which the support documentation is written for — did show all BTM-registered apps in this System Settings page, but non-enterprise Macs didn’t until macOS Golden Gate. ↩︎