The Messages app is arguably the most important app on the iPhone, and in iOS 27, Apple is adding some useful improvements. There are some quality-of-life fixes for longtime issues, and new AI features.

Contextual Suggestions

Messages uses Apple Intelligence to surface one-tap suggestions based on what's being discussed in a conversation. If someone you're talking to asks you to send a photo, Messages recognizes the request and suggests adding the relevant photos.


If someone invites you to an event, Messages offers to add it to your Calendar. If someone asks you to pick something up at the store, Messages will suggest adding it to your Reminders list or to the Notes app.

Messages will also suggest Genmoji you might want to generate based on the conversation.

Write With Siri

There's a systemwide Write with Siri feature that's available in the Messages app. ‌Siri‌ AI can compose entire messages for you, proofread your messages, or give you feedback.


There is a dedicated Write with ‌Siri‌ button above the keyboard where you can get to the new ‌Siri‌ tools.

Smart Reply suggestions also use your writing style so they sound more natural.

Siri AI

‌Siri‌ AI integrates with the Messages app. You can ask ‌Siri‌ to find images, videos, links, addresses, phone numbers, and more in the Messages app, plus ‌Siri‌ can draft message replies and act on info in the Messages app, like adding an event date to the Calendar app.

Drawing

You can use a new Drawing interface in the Messages app to send little sketches and notes. It's available when tapping on the "+" button next to the text bar in a conversation.


The drawing interface is the same as the Markup tool available in apps like Photos and Notes. There's a marker, a pen, a highlighter, and an eraser, plus options to use the tools in any color.

Accidental Recordings

The ‌iOS 27‌ Messages app includes an option to remove the audio button from the text bar. In the Messages section of the Settings app, you can navigate to Show in Text Field to set the right-side button to Record Audio, Start Dictation, or None.


Setting the option to none will prevent you from accidentally recording a voice message with an errant tap.

Tapbacks

Messages now consolidates notifications for Tapback responses, which is useful when you're in a group thread and tons of people are sending emojis.

Sending

When a message fails to send in ‌iOS 27‌, your iPhone will automatically retry sending it. In iOS 26, you get an alert if a message fails, but you have to manually tap to try to send it again.

The Messages app also supports continuous sending of photos, videos, and texts. If you're sending someone a photo and your connection is poor or you go offline, it automatically resumes sending when the connection is restored.

Syncing

Cross-device syncing is improved in ‌iOS 27‌, so conversations will sync more reliably across the iPhone, iPad, Mac, Apple Watch, and Vision Pro. Fewer messages will fail to sync and display properly when you switch between devices.

Messages, read states, reactions, and attachments sync faster.

When you take a photo and then go to the Messages app to send it, you're more likely to see it right away. Apple says the iPhone is faster to add recent camera captures to Messages.

Search

Apple improved its search index across apps like Mail and Messages, so you're more likely to find what you're looking for.

You can also search for conversations in Messages using a phone number or a contact's nickname.

Offloaded Media

If you have photos or videos that have been offloaded to save storage space, you can still find them in the ‌iOS 27‌ Messages app when you search.

Thumbnails are also displayed for offloaded media instead of a placeholder image, so you know what's there.

RCS

In RCS message conversations with Android users, you can now long press on a message to send an inline reply. ‌iOS 27‌ also shows reaction emoji directly on images and videos, a feature that was not functioning in ‌iOS 26‌.

Genmoji

Apple revamped ‌Genmoji‌ creation in ‌iOS 27‌. There's an updated interface for describing an emoji that you want, with options to start from an existing emoji, choose an image from ‌Photos‌, or select a person.


After uploading an image or using a text phrase to create a ‌‌Genmoji‌‌, there's a new "Describe a change" interface for making updates to what you've already created. You can iterate on designs and change individual elements of a ‌‌Genmoji‌‌, tweaking colors and objects. Each change uses the old base instead of regenerating a new ‌‌Genmoji‌‌.

With a series of requests, it is possible to make a complicated, multi-element ‌‌Genmoji‌‌. Along with describing changes, you can add in additional emoji, and the interface makes suggestions on what to do next.

‌‌Genmoji‌‌ output is more consistent, and ‌‌Genmoji‌‌ look more like real emoji with a 3D, cartoonish style by default. There is now an option to change style, so if you don't want the cartoonish look, you can ask for something else like a drawing or a sketch.

Large Conversations

Large conversations load and scroll faster in ‌iOS 27‌, according to Apple.

Read More

More on all of the features in ‌iOS 27‌ can be found in our iOS 27 guide.
This article, "iOS 27: 13 Ways the Messages App is Smarter and Less Annoying" first appeared on MacRumors.com

Discuss this article in our forums

A new release of our linl package for writing LaTeX letters with (R)markdown is now on CRAN. linl makes it easy to write letters in markdown, with some extra bells and whistles thanks to some cleverness chiefly by Aaron.

This version is mostly maintenance: updates to the continuous integration setup, as well as updates to packaging including use of Authors@R in DESCRIPTION. No functional changes, no new code, or new features.

The NEWS entry follows:

Changes in linl version 0.0.6 (2026-08-26)

• Several updates to continuous integration and testing

• Switch to Authors@R in DESCRIPTION

Courtesy of CRANberries, there is a comparison to the previous release. For questions or comments use the issue tracker off the GitHub repo.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. If you like this or other open-source work I do, you can sponsor me at GitHub.

The Debian LTS Team, funded by Freexian’s Debian LTS offering, is pleased to report its activities for July.

Activity summary

During the month of July, 23 contributors have been paid to work on Debian LTS (links to individual contributor reports are located below).

The team released 52 DLAs fixing 2159 CVEs.

In July, the Debian Stable Release Managers published the last point release of Debian 12 (“bookworm”), after which the Debian LTS team took full responsibility of Debian 12. This completes the handover from the Security Team, that took place in June. This also marks the second month in a row where the Debian LTS has been focusing on two simultaneous Debian releases.

Other than Debian 12, the team is maintaining Debian 11 (“bullseye”), which will reach the end of its Long Term Support on 31 August 2026. After that date, Freexian will continue the security support under the Extended LTS offer.

The team published several notable updates:

• jq (DLA 4662-1 and DLA 4661-1) prepared by Andreas Henriksson in collaboration with Jochen Sprickerhof, addressing multiple vulnerabilities.
• Several updates for the different linux supported versions prepared by Ben Hutchings, in collaboration with Emilio Pozuelo Monfort. Other than the regular security advisories: DLA 4664-1, DLA 4665-1, DLA 4671-1, DLA 4688-1, and DLA 4700-1, Ben started preparing packages of 6.12 via bookworm-backports.
• nginx (DLA 4667-1), updated for bookworm by Carlos Henrique Lima Melara, as a follow up of the bullseye update (DLA 4660-1), that was prepared in June.
• grub2/bullseye (DLA 4685-1), prepared by Emilio. Other than addressing several security issues, this DLA was needed for being able to update the shim boot loader.
• samba (DLA 4692-1), uploaded by Markus Koschany, to fix several security flaws in bullseye, including issues that could yield to remote code execution.
• imagemagick (DLA 4680-1 and DLA 4696-1), prepared by Bastien Roucariès, addressing several issues that could lead to denial of service, information disclosure or potentially arbitrary code execution in some scenarios.
• poppler (DLA 4709-1), by Guilhem Moulin, fixing several vulnerabilities.
• nss (DLA-4694-1), by Jochen, fixing flaws that may result in or denial of service or potentially the execution of arbitrary code.

Contributions from outside the LTS Team:

• Guillem Jover prepared the dpkg update, released as DLA-4673-1 by Arnaud.
• Nicholas Guriev released a rlottie security update (DLA 4675-1)
• Bernhard Schmidt prepared the openvpn updates for both bookworm and trixie, released as DLA 4666-1.

The LTS Team has also contributed with updates to the latest Debian releases:

• Bastien also proposed two updates for imagemagick. The first one released as DSA 6383-1, and the second as a trixie point update proposal (#1142554).
• python-httplib2 by Emmanuel Arias, and released by the security team as DSA 6441-1 in August.
• hplip (DSA 6402-1), prepared by Thorsten Alteholz, to address privilege escalation and arbitrary code execution related flaws.
• libnfs trixie update (#1142351), by Thorsten
• patool update for trixie #1141607, by Abhijith PA

Other contributions:

Besides the work on security updates, different documentation and tooling changes were needed, especially in the context of the Debian 12 handover. This work was mainly done by Sylvain Beucler.

Individual Debian LTS contributor reports

• Abhijith PA
• Andreas Henriksson
• Andrej Shadura
• Arnaud Rebillout
• Bastien Roucariès
• Ben Hutchings
• Carlos Henrique Lima Melara
• Chris Lamb
• Daniel Leidert
• Emmanuel Arias
• Emilio Pozuelo Monfort
• Guilhem Moulin
• Jochen Sprickerhof
• Lee Garrett
• Lucas Kanashiro
• Lukas Märdian
• Markus Koschany
• Paride Legovini
• Santiago Ruano Rincón
• Sylvain Beucler
• Thorsten Alteholz
• Tobias Frost
• Utkarsh Gupta

Thanks to our sponsors

Sponsors that joined recently are in bold.

• Platinum sponsors:
  • Toshiba Corporation (for 130 months)
  • Civil Infrastructure Platform (CIP) (for 98 months)
  • VyOS Inc (for 63 months)
• Gold sponsors:
  • F. Hoffmann-La Roche AG (for 141 months)
  • CONET Deutschland GmbH (for 124 months)
  • University of Oxford (for 81 months)
  • EDF SA (for 52 months)
  • Dataport AöR (for 27 months)
  • CERN (for 25 months)
• Silver sponsors:
  • Domeneshop AS (for 145 months)
  • Nantes Métropole (for 139 months)
  • Akamai - Linode (for 135 months)
  • Univention GmbH (for 131 months)
  • Université Jean Monnet de St Etienne (for 131 months)
  • Ribbon Communications, Inc. (for 125 months)
  • Exonet B.V. (for 115 months)
  • Leibniz Rechenzentrum (for 109 months)
  • Ministère de l’Europe et des Affaires Étrangères (for 93 months)
  • Dinahosting SL (for 80 months)
  • Upsun Formerly Platform.sh (for 75 months)
  • Moxa Inc. (for 69 months)
  • sipgate GmbH (for 66 months)
  • OVH US LLC (for 64 months)
  • Tilburg University (for 64 months)
  • GSI Helmholtzzentrum für Schwerionenforschung GmbH (for 56 months)
  • THINline s.r.o. (for 28 months)
  • Copenhagen Airports A/S (for 22 months)
  • Conseil Départemental de l’Isère (for 8 months)
  • CEOS GmbH
• Bronze sponsors:
  • Seznam.cz, a.s. (for 146 months)
  • Evolix (for 145 months)
  • Linuxhotel GmbH (for 143 months)
  • Intevation GmbH (for 142 months)
  • WAAYS (for 141 months)
  • Megaspace Internet Services GmbH (for 140 months)
  • Greenbone AG (for 139 months)
  • NUMLOG (for 139 months)
  • WinGo AG (for 138 months)
  • Entr’ouvert (for 130 months)
  • Adfinis AG (for 127 months)
  • Plat’Home (for 124 months)
  • Laboratoire LEGI - UMR 5519 / CNRS (for 122 months)
  • Tesorion (for 122 months)
  • Bearstech (for 114 months)
  • LiHAS (for 114 months)
  • Catalyst IT Ltd (for 108 months)
  • Demarcq SAS (for 102 months)
  • Université Grenoble Alpes (for 88 months)
  • TouchWeb SAS (for 80 months)
  • SPiN AG (for 77 months)
  • CoreFiling (for 73 months)
  • Observatoire des Sciences de l’Univers de Grenoble (for 65 months)
  • Tem Innovations GmbH (for 59 months)
  • WordFinder.pro (for 59 months)
  • CNRS DT INSU Résif (for 58 months)
  • Soliton Systems K.K. (for 53 months)
  • Alter Way (for 51 months)
  • SOBIS Software GmbH (for 25 months)
  • Tuxera Inc. (for 17 months)
  • OPM-OP AS (for 8 months)
  • LU-CIX Management G.I.E.

Following this morning's announcement of the iPhone event that will take place on September 9, Apple added an event placeholder to its YouTube channel.


Apple will stream the "Surprise and Shine" event on YouTube, on the Apple Events website, and in the Apple TV app wherever it is available.

YouTube users can click on the video placeholder and choose the "Notify me" option to get a notification when the livestream is up. Apple also added an event countdown.

If YouTube has your location information, the notification and countdown will be displayed in your local time, so it's a good way to double-check when you will need to tune in to watch the event.

Apple's event will start at 10 a.m. Pacific Time, which is 1 p.m. Eastern Time, 6 p.m. British Summer Time, 7 p.m. Central European Summer Time, and 10:30 p.m. India Standard Time, 1 a.m. China Standard Time (Thursday, September 10), 1 a.m. Australian Western Standard Time, 3 a.m. Australian Eastern Standard Time, and 5 a.m. in New Zealand Standard Time.

You're also able to add the event to your calendar using the Apple Events website.

For those who are unable to watch the livestream on September 9, MacRumors will have full event coverage on MacRumors.com and the MacRumorsLive X (formerly Twitter) account so you can follow along with the announcements.
This article, "Apple Event Livestream: Where and When to Watch on September 9" first appeared on MacRumors.com

Discuss this article in our forums

If you dial Apple Support's phone number 1-800-APL-CARE from the U.S. or Canada, there is a good chance that an AI-powered support assistant will now answer.


While many of Apple's phone systems already had a basic automated operator, the Apple Support phone line is now using a voice powered by generative AI. The assistant can answer questions and troubleshoot issues with your Apple products, with it able to provide detailed step-by-step instructions. The assistant can still transfer you to a human advisor if it cannot answer a specific question, or if you ask to speak to a real person.

An anonymous tipster claimed that Apple has partnered with Sierra AI on the virtual assistant, but we have not independently confirmed this information.

A concern is that Apple's increasing use of AI for customer service may result in a reduction in human advisors, through fewer job listings and/or layoffs.

It appears that the AI assistant for Apple Support phone calls went live earlier this year. Apple has also been testing a new AI-powered chatbot in its Apple Support app, through a new "Ask" tab that has rolled out to a limited number of users.
This article, "1-800-APL-CARE Now Connects You With AI Assistant Instead of Human" first appeared on MacRumors.com

Discuss this article in our forums

Images claiming to show the inside of an Apple-made AI server, believed to power Private Cloud Compute, surfaced online this week, offering a first look at how the hardware is laid out.


The images were shared earlier this week by the X user @hsuchingpo, an account with a track record of Apple hardware and prototype leaks. Captioned simply "Apple M5 Server," the leaker said the hardware is "most likely" Apple's Private Cloud Compute (PCC) system. Private Cloud Compute is Apple's system for offloading demanding Apple Intelligence requests from a device to the cloud while keeping user data from being stored or shared.

The images show a rack mountable chassis packed with compute modules. Each one has a separate circuit board mounted on a carrier bar and secured with blue and red ejector latches for quick removal. The boards connect into a backplane through gold edge connectors, with heatsinks running through the center of each column and a service diagram taped inside the lid mapping out numbered board positions and Apple part numbers.



Each single chassis holds up to 32 boards. The layout appears to be built around airflow, with heatsinks running down the center and fans at both ends.

One close-up image shows a board with its thermal paste wiped away, revealing an Apple logo printed directly onto a chip package, appearing to confirm the hardware uses Apple's own silicon. Two NAND flash chips sit nearby.

The leaker added that the server cannot manage its own workload independently. In replies, they explained that the hardware "requires use in conjunction with Mac Studio (serves the function of software control)" and apart from Apple, no one can activate it, since doing so needs both a physical control machine and a control account.

Apple produces AI servers domestically at a facility in Houston, Texas, as part of a $600 billion U.S. investment commitment.
This article, "Leaked Images Give First Look at Apple's AI Servers" first appeared on MacRumors.com

Discuss this article in our forums

Apple plans to hold its annual iPhone-centric event on Wednesday, September 9 at the Apple Park campus in Cupertino, California. The event will begin at 10:00 a.m. Pacific Time, with select members of the media invited to attend.


The September 2026 iPhone event will see Apple announce the iPhone 18 Pro, the ‌iPhone 18 Pro‌ Max, and the foldable iPhone, which could be called the iPhone Ultra. There won't be an iPhone 18 this year because Apple is waiting until spring 2027 to launch the more affordable models.

All three new iPhones will use Apple's latest A20 Pro chip built on a 2nm process for speed and efficiency improvements. The ‌iPhone 18 Pro‌ models could get an upgraded variable aperture rear camera, but will otherwise be similar to the iPhone 17 Pro models. We are expecting price increases due to ongoing memory shortages.


The foldable iPhone will be Apple's first new form factor since the launch of the 2017 iPhone X. It will have a book-like design, and it will be around 5.5 inches when closed and 7.6 inches when open. It is wider than it is tall, and will be similar to using an iPad when unfolded. Since the foldable iPhone is so thin, there are compromises, and it will use Touch ID instead of Face ID and it won't have a telephoto camera.

Apple is also going to introduce the Apple Watch Series 12 and Apple Watch Ultra 4, plus there's a possibility we'll see new smart home devices, including the long-rumored home hub, a new Apple TV 4K, and an updated HomePod mini.

Along with new hardware, Apple will give us launch dates for iOS 27, iPadOS 27, macOS Golden Gate, and other upcoming software updates.

Apple plans to stream the fall event live on its website, on YouTube, and through the ‌Apple TV‌ app. For those unable to watch, MacRumors will be providing full event coverage on MacRumors.com and the MacRumorsLive Twitter account.
This article, "Apple Event Announced for September 9: 'Surprise and Shine'" first appeared on MacRumors.com

Discuss this article in our forums

The DistroWatch news feed is brought to you by TUXEDO COMPUTERS. The Haiku project has released a new development snapshot of the lightweight, single-user operating system. Haiku R1 Beta 6 introduces support for multiple web browsers and features a port of the NetBSD Virtual Machine Manager (NVMM), which provides hardware virtualization on x86_64 machines. "A few months after the....

As a Debian developer, I have had to cast a vote for the General Resolution named LLM usage in Debian (progress report here). This was not an easy task for me…

It’s a good thing that the vote is secret so that people are not scared of voting according to their own beliefs. I have Debian friends on the whole spectrum of opinions that are represented here, and I hesitated twice on sharing my own thoughts for fear of alienating my relationship with them. But in the end, we all make efforts to respect the opinions of those who are not thinking like us, and it’s precisely that willingness to work together towards a solution that is acceptable by the majority that makes Debian so strong. So here’s the train of thoughts that I followed to cast my vote.

The difficulty for me was to reconcile the political statement that I want to make and my desire for this vote to not be (too) divisive for the Debian community, and to make sure we are not putting off newcomers with choices that might be hard to stand by in the long term.

So let’s be clear : if I had a magical wand to make AI and LLM disappear, I would use it for that purpose, since at this point in time I don’t believe that the benefits outweigh the costs that the AI race is inflicting on us. If I were a political decision-maker, I would forbid the construction of new data centers unless they also build renewable energy infrastructure to cover for their additional energy consumption. I would also legislate so that AI companies have to document what material they used to train their models, and I would forbid scraping for that purpose, and build ways for those companies to buy copies of properly-sourced training data. That is to say, I don’t like the way LLM are built by the players in that market, I’m pretty scared of the ecological impact of what those players are doing, and I’m certainly worried about the long term effect that LLM will have on society as a whole.

Nevertheless what brought me to Debian is the ability to experiment and contribute to something useful with cool technologies, and as a computer scientist, the potential of LLM done right is hard to ignore. Given what we have seen already, I expect that LLM will empower (a part of) the next generation to learn IT, computing and even Debian packaging. Completely refusing the use of LLM is likely to make it harder for us to attract new contributors. In fact, we have already seen people inside Debian that would likely stop contributing if they are now forbidden to use LLM. I know there are likely others that will quit Debian if we accept it too, but I hope we can find a middle-ground where such persons can decide that LLM are not welcome in the small corner of Debian that they are in charge of…

In the end, I decided that answering clearly the question “Shall we accept LLM contributions ?” was more important than making the political statement about the current state of affairs in the AI landscape, both because I believe that Debian statements have a negligible impact on policy-makers, and because historically Debian has grown by staying close to technical excellence and relatively far from politics, except when it comes to the way we handle people. And as much as I care about climate change, I don’t see how bringing this up in the context of a Debian statement is helping its cause.

More concretely, it gives the following ranking (in decreasing order of importance):

• B, D: those two choices are the clearest to express “Yes we should accept LLM contributions” and still acknowledge concerns about the way AI is built today
• F, H: those two choices do not forbid LLM usage but discourage their use and clearly voice the concerns
• E: this choice is basically the statu-quo and fails to acknowledge the concerns, but it does not forbid LLM usage
• None of the above
• G, A, C: those choices forbid LLM usage in various ways

I don’t know what option will win, but assuming that LLM-assisted contributions are allowed, I believe that it would be helpful to have further statements to clarify a few things:

• Even if Debian as a whole doesn’t want to ban LLM-assisted contributions, each maintainer or each team shall be free to forbid LLM assisted contributions in the parts of Debian that they are maintaining
• We should discourage usage of LLM provided by players with unethical behaviors (not sure if there are good players but well…)
  

The iPhone 17 retained its position as the world's best selling smartphone in the second quarter of 2026, according to Counterpoint Research, with the iPhone 17 Pro Max and ‌iPhone 17 Pro‌ holding the next two spots on the firm's quarterly ranking.


Data from Counterpoint's Global Handset Model Sales Tracker places the standard ‌iPhone 17‌ at 6% of global smartphone unit sales for the second quarter, building on the top spot it already held in the first quarter. Apple and Samsung filled all ten of the top spots, with a combined 26% share of worldwide unit sales that was two points higher than a year earlier.

Counterpoint tied the growth to regional strength outside Apple's usual core markets, with gains in India, Japan, and the Middle East and Africa, plus a near doubling of sales in South Korea.

The base iPhone 17 model remained the best selling smartphone during the quarter, supported by significant upgrades that narrowed the gap with the Pro models. Meanwhile, strong premium demand drove the iPhone 17 Pro and Pro Max sales, aided by trade in programs, promotions, and accessible financing options that improved affordability.

Rising smartphone prices across the industry also nudged buyers toward more expensive pricing tiers generally, which largely worked in Apple's favor given its lineup skews upmarket already.



The iPhone 17e took seventh place following its refresh that added MagSafe and increased storage for the same price, boosted by compelling carrier deals in the U.S. and Japan. The iPhone 16 rounded out the list at number ten, leaving Apple with five of the ten slots, which is the same count and nearly the same positions Apple held a year earlier with the ‌iPhone 16‌ lineup and iPhone 15.

Samsung's best performer was the Galaxy S26 Ultra in fourth place, which is the first time that Samsung's top tier device has outsold every other Android phone in the second quarter. This is said to partly be a reflection of Samsung steering more marketing spend toward its flagship models to protect margins as costs climb. Samsung's cheaper Galaxy A phones also remained competitive further down the market.

Overall, the smartphone market shrank 11% in the quarter, a contraction Counterpoint links to the ongoing shortages and price rises that has manufacturers concentrating resources on their most profitable phones. Apple and Samsung both grew unit sales against that backdrop, a divergence that is visible in Counterpoint's regional data as well.

This article, "iPhone 17 Remains World's Best Selling Smartphone" first appeared on MacRumors.com

Discuss this article in our forums

For anyone who's experienced the lengthy post-update indexing process when trialing the beta version of iOS 27, macOS 27 Golden Gate, iPadOS 27, or visionOS 27: Apple has published a Spotlight support document that explains what's going on.

Apple says indexing can take hours or even days depending on how much data you have stored. It completes more quickly when your device is not in use, and when it is connected to power and the internet. Apple also confirms that the index is stored locally, remains private, and is not shared with Apple or any of your other devices.
This article, "Apple Explains Why Spotlight Indexing Can Take Days After an Update" first appeared on MacRumors.com

Discuss this article in our forums

Starting today, Virginia residents can add a digital version of their driver's license to the Apple Wallet app on the iPhone and Apple Watch. The digital ID can be used to display proof of identity or age at TSA checkpoints in select airports, at select businesses, and in select apps, but the number of places where it is accepted is still quite limited.


Virginia is the 15th state to implement support for Apple Wallet IDs in the U.S., joining Arizona, Arkansas, California, Colorado, Georgia, Hawaii, Illinois, Iowa, Maryland, Montana, New Mexico, North Dakota, Ohio, and West Virginia. The feature is also available in Puerto Rico, and it works with Japan's My Number Card.

Apple has been preparing for the feature to go live in North Carolina, Oklahoma, and Utah in the coming months, and years ago it announced that Connecticut, Kentucky, and Mississippi were committed to launching the feature at some point.

To set up the feature, open the Wallet app on the iPhone and tap on the plus sign in the top-right corner. Next, tap on Driver's License and ID Cards, select Virginia from the list, and follow the on-screen steps to complete the process. Then, you can present your digital ID where it is accepted without handing over your device.


You can present an Apple Wallet ID where accepted by simply tapping your iPhone or Apple Watch on an identity reader. Then, you will be able to view the specific information being requested by the TSA or business. Only after authorizing with Face ID or Touch ID is the requested identity information released from your device.

As mentioned, you do not need to unlock, show, or hand over your device to present an Apple Wallet ID in person, ensuring user privacy.

Apple Wallet IDs are generally not accepted by law enforcement, so carrying a physical ID is still legally required for traffic stops. In addition, the number of businesses that accept Apple Wallet IDs is still quite small. At least for now, Apple Wallet IDs are designed to be a convenient alternative where they are accepted.

Apple Wallet IDs are accepted at TSA checkpoints in more than 250 airports across the U.S. for domestic travel, according to Apple.
This article, "Apple Wallet Driver's License Feature Just Launched in Virginia" first appeared on MacRumors.com

Discuss this article in our forums

Apple yesterday announced the new Mac mini and Mac Studio, and pre-orders have begun on each device ahead of their September 22 launch date. We aren't currently tracking any straight cash discounts on pre-orders, but Best Buy is offering some money back in the form of Best Buy rewards when ordering.

Note: MacRumors is an affiliate partner with Best Buy. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

Specifically, My Best Buy Plus and Total members will get 2% back in the form of Best Buy rewards with the pre-order of a new Mac mini or Mac Studio. This offers runs through September 22 and is available to new and existing My Best Buy Plus and My Best Buy Total members, which starts at $29.99/year.




You can find out more about everything Apple announced yesterday in our dedicated post. This also includes the new polishing cloth and a few updates to the Magic Keyboards.

Mac mini

• M6 16GB/256GB - $899.00


• M6 16GB/512GB - $1,099.00


• M6 24GB/512GB - $1,299.00


• M5 Pro 24GB/512GB - $1,699.00

Mac Studio

• M5 Max 36GB/512GB - $2,499.00


• M5 Ultra 96GB/1TB - $5,499.00

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.

---

This article, "Pre-Ordering a Mac Mini or Mac Studio? Here's How to Get Money Back at Best Buy" first appeared on MacRumors.com

Discuss this article in our forums

Apple's new Mac mini and Mac Studio feature genlock support over USB-C, allowing users to precisely synchronize a display with a camera such as the iPhone 17 Pro, without the need for additional equipment.

It keeps multiple cameras and displays locked to the same timing signal, stopping footage from drifting out of sync during a multi-camera shoot.
This article, "New Mac Mini and Mac Studio Add Genlock Support for Camera Syncing" first appeared on MacRumors.com

Discuss this article in our forums

A new fan-made concept video imagining Apple's rumored foldable iPhone is circulating online, weeks before the company is expected to unveil the device at its annual fall keynote.


The video was created by designer Gunhoo Lee to provide a realistic look at what a brief "iPhone Ultra" reveal could look like, showcasing its 7.8-inch inner display, Touch ID authentication, mechanical hinge, dual battery system, and A20 chip. The features and specifications it depicts track closely with months of corroborated rumors.

Apple's foldable iPhone is widely expected to be announced alongside the iPhone 18 Pro and ‌iPhone 18 Pro‌ Max next month, so the real reveal video isn't far away.
This article, "Concept Video Imagines Apple's Foldable iPhone Ultra Reveal" first appeared on MacRumors.com

Discuss this article in our forums

iOS 27 tweaks some of the information and menus shown alongside the Digital ID passport feature in the Apple Wallet app.

As a result of these changes and/or a related bug, some iPhone users have seen their Digital ID deactivated on recent iOS 27 beta versions. Apple says affected users must remove the card and re-add it to the Apple Wallet app in order to use it again.
This article, "iOS 27 Tweaks Apple Wallet's U.S. Passport Feature" first appeared on MacRumors.com

Discuss this article in our forums

Apple has informed some customers who recently ordered a Mac mini with an M4 chip that has yet to ship that it will be upgrading their order to an equivalent version of the new Mac mini with an M6 chip at no additional cost.


Apple's email to customers:

Thank you for your recent Mac order. We know you're looking forward to receiving your purchase.

As you may know, Apple recently announced the new Mac mini. Since your order has yet to ship, we automatically upgraded you to the new Mac mini at no additional cost.

Given the Mac mini now starts at $899 instead of $799 in the U.S., some customers are effectively receiving a $100 discount on the new model.

One customer on social media was delighted to learn that they would not only be receiving the new Mac mini at no additional cost, but also saw their delivery estimate bumped up from October to September. On the other hand, a student emailed us to complain that their delivery estimate was pushed back by several weeks as a result of the free upgrade, as they said they need the computer imminently for their studies. The student said Apple did not offer them a choice to accept or decline the free upgrade.

Apple announced the new Mac mini with M6 and M5 Pro chip options this week, alongside a new Mac Studio with M5 Max and M5 Ultra chip options. Both new Macs are available to pre-order now ahead and launch on Tuesday, September 22.

The previous-generation Mac mini was released in November 2024.
This article, "Apple Giving Some Mac Mini Customers a Free Upgrade" first appeared on MacRumors.com

Discuss this article in our forums

Multiple new Mac Studio units can be clustered together over Thunderbolt 5 and RDMA (remote direct memory access), creating a shared memory pool for running larger AI models.

If four Mac Studio systems are clustered together, Apple says they can deliver 3x faster AI inference than running the same task on a single system, with the shared memory pool helping to load the largest and most demanding frontier-class open-weight models available today.
This article, "New Mac Studio Can Be Clustered Together for Faster AI Performance" first appeared on MacRumors.com

Discuss this article in our forums

Three out of four U.S. adults say they have no interest in buying a foldable iPhone, according to a new survey from CNET.


CNET's survey was conducted online by YouGov between August 13 and 17, polling 2,605 U.S. adults. It found that 75% are not interested in an Apple foldable device, and those who are would not pay anywhere near the rumored $2,000+ price.

Interest is low across every generation surveyed. Boomers (1946–1964) and Gen Xers (1965–1980) are the least interested, at 83% and 77% respectively, while 69% of Gen Z (1997–2012) and 68% of millennials (1981–1996) also said they would pass on a foldable iPhone. CNET noted that a quarter of respondents still expressing interest in the device is a notable proportion for what remains a niche product category.

On average, U.S. adults said they would spend just $781 on a foldable iPhone, about $48 less than the $799 starting price of the standard iPhone 17. Gen Z was willing to spend the most, an average of $919, followed by millennials at $859, Gen X at $690, and Boomers at $572. That mismatch between expected pricing and what buyers actually want to pay highlights just how far a $2,000+ device will sit outside most people's budgets.

Moreover, CNET found that 41% of U.S. adults are rethinking their upgrade plans because of anticipated price increases. Apple raised prices on most of its devices in June, citing the global memory shortage, though iPhone, Apple Watch, and AirPods prices were untouched at the time. The iPhone 18 Pro lineup is expected to cost more than its predecessor when it launches for the same reasons.

5% of respondents said they are considering buying an older or cheaper model, 4% are considering switching brands, and 4% are looking at refurbished or used phones. Another 9% said they will only upgrade if a trade-in or carrier promotion covers the cost. 22% of U.S. adults said they have no plans to upgrade anytime soon regardless of what Apple releases, ranging from 16% of millennials to 27% of Boomers.
This article, "75% of US Adults Aren't Interested in a Foldable iPhone" first appeared on MacRumors.com

Discuss this article in our forums

The DistroWatch news feed is brought to you by TUXEDO COMPUTERS. The Armbian distribution is designed for ARM development boards and a range of other machines. The project's latest release, version 26.8.3, introduces a few key change, including a new system installer. "Most releases are a long list of small improvements. This one had three larger pieces landing at....

Apple's new $9 Polishing Cloth, whose changes beyond pricing were unclear yesterday, is close in size to the original but noticeably less stiff, according to a comparison posted on X.

The user says the new material feels closer to the cloth included with Apple's nano-texture displays or the Vision Pro.
This article, "Apple's New Polishing Cloth Is Softer, Not Smaller Than Original" first appeared on MacRumors.com

Discuss this article in our forums

Apple's newly announced M6 and M5 Pro Mac mini models still support up to three external displays, but there's now support for higher refresh rates, according to Apple's published technical specifications.


Here's what's changed versus the M4 generation:

• The M6 model runs two displays at up to 6K 60Hz or 4K 165Hz, up from 144Hz on M4, plus a third display at up to 5K 60Hz over Thunderbolt or 4K 60Hz over HDMI.


• The M5 Pro model runs all three displays at up to 6K 60Hz or 4K 165Hz, up from 144Hz on the M4 Pro.


• In a two-monitor setup, the M5 Pro model's second screen now supports 5K 120Hz or 4K 200Hz, whereas the M4 Pro listed 6K 60Hz or 4K 144Hz.


• The M5 Pro model supports all three displays through one Thunderbolt port, while the M6 still supports two per port.


• Single-display support is unchanged: Both chips support 8K 60Hz, 5K 120Hz, or 4K 240Hz.

As for port compatibility around the back, the M6 model has Thunderbolt 4 supporting DisplayPort 1.4, while the M5 Pro model has Thunderbolt 5 supporting DisplayPort 2.1.

The new models were announced on Tuesday. Pre-orders are open now, with shipping and in-store availability on September 22.
This article, "New Mac Mini Bumps Display Refresh Rates to 165Hz" first appeared on MacRumors.com

Discuss this article in our forums

Images and video claiming to show circuitry from Apple's upcoming foldable iPhone are circulating on X and Chinese social media platforms.


For example, on X, the user "LusiRoy7" posted a short video alongside an image, as well as a labeled composite image identifying components, purportedly depicting the motherboard of Apple's foldable iPhone. The image appears to show the CPU, a hinge connector, a flex cable connector, and more.

Separately, on the Chinese forum Baidu, a diagram from the repair parts vendor MIJING depicts the same components. The MIJING diagram uses Chinese repair-industry component numbering. The range of images appear to show the same parts.

While the authenticity of the images is unknown, they are likely another sign that the launch of the first foldable iPhone is nearing. The device is widely expected to be announced alongside the iPhone 18 Pro models next month.

Rumors point to a 5.5-inch cover display and a 7.8-inch inner display, the A20 Pro chip, and a 5,000mAh battery, as well as a 48-megapixel dual rear camera system and 18-megapixel selfie cameras on both the outer and inner displays. The device is expected to debut with a price of at least $2,000.

This article, "Alleged Foldable iPhone Motherboard Photos Surface Online" first appeared on MacRumors.com

Discuss this article in our forums

Apple's Car Keys feature appears to be coming to supported vehicles made by Chinese maker Xiaomi Auto, based on code changes discovered by MacRumors in Apple's Wallet app backend.


Code spotted in the third developer beta of iOS 27 back in July revealed that Apple was preparing Car Keys support for Xiaomi, so this latest change suggest support is imminent.

The Chinese electronics maker has expanded into EVs with models like the SU7 sedan and YU7 SUV. Car key support would be one of the company's first tie-ins with Apple's ecosystem, despite Xiaomi being a major Apple rival in the smartphone market.

Car Keys allows an iPhone or Apple Watch with NFC capabilities to unlock a vehicle through the Wallet app. A digital version of a car key is stored in Wallet, and unlocking can be done simply by holding an Apple Watch or ‌iPhone‌ near a compatible vehicle's NFC reader.

What can be done with Car Keys may vary by car manufacturer, but at a minimum, Car Keys can be used to unlock your car, lock your car, and start your car, which are the features available with a physical key.

Apple introduced Car Keys in 2022, and car manufacturers like BMW, Rivian, Kia, and Hyundai have all implemented support for Car Keys. Apple maintains a full list of vehicles that support Car Keys on its CarPlay model availability webpage.
This article, "Apple Car Key Support Coming to Xiaomi Auto Vehicles" first appeared on MacRumors.com

Discuss this article in our forums

Apple yesterday announced the M6, its first chip built using a 2nm process and the first Apple silicon design to use three different CPU core types, but how does it compare to the M5?


Compared to the M5 chip that Apple launched in October 2025, the M6 delivers:

• Up to 1.2x faster multithreaded CPU performance.


• Nearly 30% higher peak GPU compute for AI.


• Up to 2x peak Neural Engine compute.


• 50% higher geometry rates for complex graphics.


• 10% higher unified memory bandwidth.

Apple did not publish real-world benchmarks comparing the M6 to the M5 directly, instead framing its Mac mini figures against the M4 and M1 models:

• Up to 13.5x faster LLM prompt processing in LM Studio versus ‌M1‌, and up to 4.8x faster versus the M4.


• Up to 2.3x faster spreadsheet calculations in Microsoft Excel versus ‌M1‌, and up to 1.5x faster versus the M4.


• Up to 2x faster gaming performance with ray tracing in "Cyberpunk 2077: Ultimate Edition" versus the M4.

The most significant architectural change is in the CPU. The M5 splits its ten cores into performance and efficiency groups, which is an arrangement Apple silicon has used since the ‌M1‌. The M6 adds a third tier called "super cores," which handle single-threaded workloads and which Apple says give the chip the world's fastest single-threaded performance.

The performance cores join them for heavier multithreaded work, while the efficiency cores take background tasks. The second major change is the Neural Engine, which is now doubled, with system frameworks able to use both engines at once so apps see faster model execution without any work from developers required. Other hardware changes compared to the M5 include:

Apple M5 Chip (2025)	Apple M6 Chip (2026)
Made with TSMC's third-generation 3nm process (N3P)	Made with TSMC's first-generation 2nm process (N2)
FinFET transistors	Gate-all-around nanosheet transistors
10-core CPU with 4 performance cores and 6 efficiency cores	12-core CPU with 2 super cores, 4 performance cores, and 6 efficiency cores
	Super cores dedicated to single-threaded workloads
16-core Neural Engine	Dual 16-core Neural Engine that system frameworks can run simultaneously
10-core GPU with a Neural Accelerator in every core	12-core GPU with a Neural Accelerator in every core
Enhanced shader cores	Updated shader core architecture
Second-generation dynamic caching	Third-generation dynamic caching
153 GB/s unified memory bandwidth	170 GB/s unified memory bandwidth

The move to a 2nm process is a significant technical change. Apple chips have used FinFET transistors for years, but the M6's N2 process uses a newer design with gate-all-around nanosheet transistors. This means that the transistor's gate wraps around the channel on every side, giving it better control and reducing power leakage at tiny sizes.

TSMC says N2 can offer 10–15% more speed at the same power, or 25–30% lower power use at the same speed, compared with its N3E process. But the M5 uses the newer N3P process, not N3E, so the actual M5-to-M6 improvement is likely to be smaller than those figures suggest.

The M6 is a broad but fairly modest upgrade. It improves the CPU, GPU, Neural Engine, and memory bandwidth, rather than delivering a dramatic gain in one area. The M5's biggest leap was in AI-focused GPU performance, thanks to Neural Accelerators built into each GPU core. The M6's reported 1.2× multithreaded CPU gain and 10% increase in memory bandwidth are good refinements rather than step-changes.

Single-threaded performance is the main exception. Some tasks cannot be split across many CPU cores, including code compilation, file indexing, image editing, and parts of agentic AI workflows. These tasks depend heavily on the speed of one core, and the M6's new "super cores" are designed to improve that directly. A doubled Neural Engine should also help with local AI models, while the larger GPU can process prompts faster.

For everyday work and general responsiveness, the difference will probably be difficult to notice. The M5 already has more than enough performance for normal use, so there is little general-purpose reason to upgrade from an M5 to an M6.

The more consequential question for buyers is where the M6 sits in the lineup, because it is not part of a straightforward replacement for the M5 family. Bloomberg's Mark Gurman reported in July that Apple has no plans to release "M6 Pro," "M6 Max," or "M6 Ultra" chips, having decided to accelerate development of the AI-optimized M7 generation. The new Mac mini illustrates this point, since its higher-specification configuration uses the M5 Pro. That makes the M6 the first of its kind to not have higher-end variants, and it means the M5 Pro, M5 Max, and M5 Ultra remain the chips to buy if you need anything more than base-tier performance.

The dual Neural Engine and 2nm process give the M6 headroom for on-device AI tools that grow more demanding, but the M7 is expected in the first half of 2027 with neural processing upgrades that prompted Apple to cut the M6 lineup short. If your workloads are AI-heavy enough that the M6's gains matter to you, the chip designed for exactly that is less than a year away.

The M6 chip is currently available only in the new Mac mini, which starts at $899 and ships on September 22. M6 versions of the MacBook Pro and iMac are expected to be announced in October. The M5 chip remains available in the 14-inch ‌MacBook Pro‌, the latest iPad Pro, and the Apple Vision Pro, with M5 Pro and M5 Max in the higher-end ‌MacBook Pro‌ models and M5 Max and M5 Ultra in the new Mac Studio.
This article, "M5 vs. M6 Chip Buyer's Guide: How Much Better Really Is M6?" first appeared on MacRumors.com

Discuss this article in our forums

I’m on a VPN setup with several friends that, obviously, includes a VoIP network. I also have an old magicJack adapter and a deep and abiding need to use hardware in ways I should not. There was obvious synergy here.

Plugging in the magicJack gives a USB vendor id of 0x06e6, which belonged to a company called TigerJet who made a range of chips for hooking up phones to computers, either via USB or PCI. Some more digging suggested that it was a 580 part, and someone had conveniently uploaded some reference code and datasheets, so figuring out how to talk to the chip wasn’t terribly difficult. Once configured it simply sends HID events whenever a user hits a phone key or changes the hook state, and otherwise exposes a USB audio device that can be spoken to using the stock kernel driver. It also has the ability to generate dial tone and assert ring signal, giving a full traditional phone experience.

So you’d think this would be a super easy project, but I’d made things harder for myself by deciding I wanted to tie directly into Asterisk rather than just smashing an existing SIP stack onto the device. Asterisk uses channels to talk to devices, and channels end up as compiled C code that Asterisk can load dynamically. I didn’t want to have to deal with the pain of compiling stuff and matching ABIs and everything so writing a new channel from scratch was unappealing. Fortunately, the websocket channel is available in recent versions of Asterisk and provides a convenient way to get audio in and out, but that still leaves the job of handling incoming and outgoing calls. That’s handled with the Asterisk Rest Interface, which can initiate a call or respond to an incoming one and bridge various channels together to produce a bidirectional audio stream. There’s a convenient async Python library that handles the low level protocol.

Code for all this is here¹, and works for my use case, but I should really abstract out the asterisk side and the magicJack side to make it easier to adapt to other devices. That’s a job for later, though. For now, you get this:

Also in this series:

• The Four Horsemen of the LLM Apocalypse
• The people vs the AI overlords

After ranting and railing about LLMs or "AI" as the optimists (or accelerationists?) call it, I figured it might be important to be a little more honest about my use of LLMs and how I think about it more practically in the world.

The Debian vote context

This is not a coming out. I am not using LLMs on a daily basis, and this blog is, again, written out of my cold dead hands in a dying world, with over-engineered hardware and (to a certain extent, hi Emacs!) software, powered by 100% green energy built on stolen land.

There is a vote going on in Debian. If you're unfamiliar with it, you can catch up at LWN. So far I've essentially said "LLM is bad" which is not a very balanced or useful opinion. Obviously, people are using LLMs, sometimes unknowing or unwillingly, and we need to take that into account. Furthermore, there has been many different blog posts on Debian planet about this. Some that I found balanced, good summaries, even if I didn't fully agree with them, at least some did the basic civil service of being short. But others were just not only Wrong but also so long that I couldn't finish that I just had to write something.¹

This is not an explanation of the ballots, nor how I will vote. This vote is Debian's failure of framing that debate in a reasonable way: we have 8 options on the ballot with many duplicates. We have failed to do the hard work of summarizing and aggregating options into a meaningful set. I doubt the final vote will represent a readable position we can rally around.

I have not read the two months of debates on the topic either. Normally, before voting, I take a cursory look at the debate to see points of view I might have missed. But in this case, it will just make me sad, add noise, and I'm already pretty sure on where I stand on this.

So let me describe how I use LLMs and how I think they fit in our work, as computer engineers and hobbyists.

My LLM use

Debian Packaging

An astute reader has pointed out that I maintain a package in Debian made to use Anthropic. It's actually multiple packages:

• llm: a CLI utility and Python library for interacting with Large Language Models, with OpenAI as its default API backend
• llm-anthropic: a plugin for llm which allows me to talk to Anthropic's API instead of OpenAI
• anthropic-sdk-python: the SDK llm-anthropic requires to do its work

As I previously explained in response, I am not entirely comfortable with this work: it's a compromise. In fact, I first uploaded llm to the contrib section of Debian, where we keep software that depends on other non-free software, but I was told that, since yt-dlp was in main, llm belonged there as well.

So I moved it to main, alongside similarly controversial tools like llama.cpp or the python-openai library.

OpenAI and Anthropic usage

An important part of my work is technology watch. I keep tabs on thousands of (new and old) software projects, follow news, and generally try to keep my skills up to date. It's a pretty impossible race, especially as I grow older, but I still think I'm doing the right choices in my job.

Testing large language models is part of that work. At first, I was using ChatGPT's web interface, but it was annoying to copy-paste things into a browser, so I looked for different interfaces.

For a while I tried gptel, a "simple, extensible LLM client for Emacs" but I found it kind of terrifying. Giving a LLM control over an Emacs buffer seems like a security nightmare, so I stopped doing that.

So I use the llm command-line tool to talk to Anthropic's API. I started that in the summer of 2025, when I bought 20$USD of API credits. Before that, I paid for a ChatGPT subscription and then OpenAI credits, which expired and sent me over to Anthropic, which seemed then to have better ethics.

As it turns out, Anthropic is also happy to work for the US military (which is a big red line for me). Anthropic also won't let you talk about the genocide in Gaza, it is destroying physical books, and is blackmailing us to use their product for security coverage.

Needless to say, Anthropic and "Claude" are not my friends, but they seem like the lesser evil in current "frontier models". So I have renewed, a couple of weeks ago, another 20$USD of API credits with Anthropic.

Actual prompts and responses

So what does 20$ give you at Anthropic anyways? What am I using LLMs for and how?

The neat thing with llm is that everything is logged in a sqlite database, so there are some answers that are easy to get:

> llm logs status
Logging is ON for all prompts
Found log database at /home/anarcat/.config/io.datasette.llm/logs.db
Number of threads logged:   7
Number of turns logged:     12
Number of legacy conversations: 543
Number of legacy responses: 970
Database file size:         9.61MB

That is 10MB of logs, with about a thousand prompts.

My logs go back to 2024-03-07, a little over two years ago, and include a mix of Anthropic and OpenAI responses. I used it more in 2024 than 2025, and if the trend continues, I will have used it less in 2026 again:

> llm logs list -n 0  --json | jq -r .[].datetime_utc | sed 's/-.*//' | sort | uniq -c 
    527 2024
    357 2025
     98 2026

It looks like about 10 prompts per month right now, down from a peak of about 60 per month in 2024. It's pretty difficult to analyze those actual logs to get more patterns and I won't run the prompts through a model again to process them.

How I'm using models now

At first, I was using it partly for benchmarking model's capabilities, like Simon Willison does with his pelicans, clearly not trusting its output. But I was impressed by the capacities of the Claude Opus 4.5 model when it wrote this script in January. Impressed, but also scared: it's the first time I felt I could delegate the entirety of my programming to a model. Just run the code, if it works, it works, right?

So what do I use it now? As an example, here are the 10 last prompts in my history:

1. there is now Claude 5, and a fable model, maybe you know about it?
2. impress me
3. not impressive, i already know all of this
4. chat
5. in postfix, i have a 300k mailing that happens regularly here. normally, it delivers within about...
6. is there a way i could have drained the maildrop queue faster without removing the milter?
7. the problem was that rspamd was timing out on the FUZZY_CALLBACK check. how do i disable that?
8. how do i disable all spam checks? i just want rspamd to add dkim signatures
9. how do the default_destination_concurrency_limit and initial_destination_concurrency settings int...
10. mic check

The first one was me trying to confirm which model I am using, which is not always obvious when going through the whole llm stack I've been using. The following two are an attempt at seeing what the model is capable of and I was "not impressed", to which Claude answered that I have a "high bar", which, fair enough.

The chat is me failing to use a command line, which shows that perhaps I need to readjust that "high bar", again.

The next five are a rather embarrassing debacle in a large Postfix mailing that went sideways, and where I couldn't find an actual Postfix expert of my level to help. The fabled Claude Fable 5 answered rather correctly, but dangerously, that I could empty the queue by disabling the non_smtpd_milters. What Fable (and myself) did not realize is that the milter was also adding DKIM signatures, so while the mailing was expedited, it was done without those precious signatures, which got us promptly blocked at Gmail. We have recovered since, and, thanks to the model and reading the Postfix manual for the hundredth time, that pickup(8) is single-threaded and that we needed to review the architecture of that mailing (and our spam filters) a bit. Many tickets ensued.

The last one is a test I did to make sure my last uploads of llm-anthropic and its dependency worked correctly.

Note that the above excludes 5 questions I asked Anthropic while writing this article, where I asked for synonyms and "what nanometer scale are arduino processors built from? how is an arduino CPU printed?", a question which Wikipedia furiously evades providing a good answer.

Those prompts are pretty typical of my LLM use: I'm testing the models to see if they work at all, but also, out of desperation, I fire off a prompt after I fire off questions to colleagues or search engines (in that order). It's often weird edge cases like the Prometheus query language, Python's matplotlib, LaTeX, Elisp, optimizations, and so on.

I use models for translation a lot. Being fully bilingual, it is common for me to think of a word in French or English and fail to find exactly the right word for that in the other language. Models help with that, and are also useful to find synonyms. Those are low-token uses that seem pretty innocuous to me, but I realize the irony of this after writing about the tower of Babel.

What I am not using models for

I am not using models to write prose.

I am not using models to read prose. If it's generated with LLMs, I stop reading.

I am not using models to write code, with the exception of that single Python script above.

I am generally not using models to review code, with exceptions. If I get stuck on a hard problem, I might feed a piece of code to the model. I repeatedly fed asncounter into Claude to try to fix a performance regression I had introduced. It found micro-optimizations that taught me a thing or two about Python's internal implementations, but overall, it was mostly a waste of time. This was in June 2025, so perhaps now models would fare better. I have not tried again.

I am not using LLMs to do Debian packaging. When I can, I manually review the diffs of packages I upload into Debian, still, by hand.

I do this for the reasons outlined in The Four Horsemen of the LLM Apocalypse, because I refuse to be complicit in the:

1. aggressive and illegal scraping of the servers I steward
2. world-wide computer hardware shortage (making it, by the way, nearly impossible to run presumably clean local models) and the attack on our job conditions (also discussed in The people vs the AI overlords)
3. death of copyright and free software
4. complication and enshifitication of everything, and the destruction of our communities
5. the imperialist Nerd Reich that wants to take over the world

Like I reluctantly use Intel computers, I do fire off a prompt. But I still hold on to the dream that we can build communities of practice that hold human knowledge collectively and not offload that as a utility to some megalomaniac billionaire.

Their LLM use I am forced into

So that's me. Clearly, I'm going against the grain here. Everywhere I look, I see LLM-generated code and projects. Slop and botnets have flooded the web.

I use Wadamesh, clearly vibe-coded, because it's the best graphical interface for MeshCore that runs on portable devices. I wish it was made by a human, in a community I could participate in, but it isn't, and I don't.

I package the above llm toolset, which is more and more vibe-coded, but I still review the diffs. And I have to say: I trust Simon here. The code is verbose as hell, feels overengineered, and llm feels slow, but it generally works, and Simon is still at the gate.

The Anthropic SDK is another thing entirely. The 0.91.0 to 0.120 upload, for example, was nuts:

 806 files changed, 72281 insertions(+), 1478 deletions(-)

I explicitly did not review that entire diff. It feels like there's a lot of garbage there to just have a shim between a proprietary API and Python. But this is the hand I've been dealt.

Larger projects LLM use

LLMs are being used in the Linux kernel, Firefox, rsync, Rust, and other places. I don't feel good about this, particularly in Rust, but they at least made a decent policy. I am glad GCC made a policy against LLM contributions and I support the human Emacs project.

We need to have a set of foundational tools that are "clean" in the sense that they are built upon a community of people that understand how they are built.

Maybe that's naive or even impossible. The Linux kernel and GCC, in particular, are massive projects that have long grown past the scale of a single person's understanding. But the theory was that a community of humans can understand collectively.

Now we seem to be throwing up our hands and giving up on that community. That LLMs will just fix the problem, whatever it is. But we're all just one rug pull away from being completely incapable of managing those projects. The argument there is that we'll just switch to local models, but no one is actually doing that. All I see is people use local models as a corner case (for privacy) or as in theory, but in reality, everyone uses the centralized frontier models right now. We just can't fallback.

We're in the same situation we were, a decade or two ago, when Microsoft decided it would kill free office alternatives by making Office free for non-profits. It worked: thousands, if not millions of schools, community groups and individuals stopped looking for alternatives (including free software but also "piracy") for Office and embraced what seemed like a generous offer.

Now Microsoft pulled the plug and Over 170,000 Nonprofits Lost All Their Data.

I'm afraid the rug pull on LLMs will be much worse: never mind that Linus won't be able to use his tireless helper to fix obscure kernel bugs; we're looking at a collapse of the economy so large that we are already talking about bailing out the companies responsible.

In a sense, the most striking thing about the Debian vote is it has actually no option to completely refuse upstream LLM contributions. It seems the community has taken it for granted that it's now impossible to build Debian entirely without LLMs. We lost the battle even without a fight, it seems.

A plea for small

If it has really become impossible for us to manage the complexity we have built, maybe it's time to stop and think about what we're doing in the first place. We're struggling to even bootstrap our current toolchain!

This is one of the things I like the most about working on the mesh: it's low tech, small Arduino devices that is built with decades-old semiconductor processes that is understandable by human beings.

Maybe the answer lies more in single-purpose devices like those communicators and simpler multi-purpose computers than what we have now, which is what the permacomputing movement is about.

Small is beautiful, let's scale it down.

BenQ today announced the launch of the iScreenBar, which it says is the first monitor light bar built for the iMac.


The iScreenBar is meant to look like an extension of the ‌iMac‌, offering desk lighting that does not clash with the ‌iMac‌'s design or block the camera or microphone. It's made from aluminum and lights an area of 33.5 by 19.7 inches at 500 lux. A glass touch panel is available for adjustments, and there's also a macOS app for tweaking brightness, color temperature, and light mode.

It turns on and off using presence detection and can also activate when the ‌iMac‌ wakes, plus it dims or brightens as the lighting in the room changes. In video call mode, evens lighting and reduces glare on the face for a better image when the camera is in use.


The light bar mounts to the back of the ‌iMac‌ using a magnetic clamp, and it comes with a magnetic cable clip to hide cords. BenQ says asymmetric optics light the user's desk without shining on the screen or in the eyes to reduce glare and preserve color accuracy. It is powered by the ‌iMac‌'s USB-C port, which also allows for automatic camera mode activation.

The iScreenBar can be purchased from BenQ or from Amazon for $189. It is compatible with the 24-inch ‌iMac‌ (2021 and later).
This article, "BenQ's New iScreenBar is a Monitor Light Bar Made for the iMac" first appeared on MacRumors.com

Discuss this article in our forums

My hobby: following GOV.UK to look for interesting announcements. Today was an update on the MOD’s Rapid AI Delivery Taskforce which was previously announced in June during London Tech Week.

I like this line: “Success is measured in operational advantage delivered, not technology demonstrated.” To me it recalls “Working software is the primary measure of progress” from Principles behind the Agile Manifesto – if you understand “working” to mean “working in production”. Which I do.

For anyone interested in suggesting ideas to the taskforce, the four operational challenge areas include:

• Understanding and decision advantage
• Electromagnetic and information advantage
• Planning and automation
• Autonomous systems

Yesterday’s announcement of UK access to Ukraine’s Avengers AI Labs database seems incredibly relevant to that last point.

Machine assistance for handling and interpreting huge volumes of data would probably benefit decision advantage and interpretation of a crowded EM spectrum, but this is hopefully(?) more than just LLMs. Of course, there’s more to AI than large language models… right?

I worry that “planning and automation” might amount to “generating large amounts of text faster”. Nothing could possibly go wrong with this.

Ads in Apple Maps are live in the United States and Canada, which means iPhone and iPad users will see them when searching for locations and when viewing Suggested Places.


Apple announced plans to bring ads to the Maps app in March, and said the feature would roll out starting in late summer. Apple confirmed to 9to5Mac that ads began showing up for users over the last few days, with ad placement to ramp up in the coming weeks.

Ads are shown in Suggested Places when initiating a search and after searching if a company has bought an ad for a keyword you're looking for. Ads are shown at the top of search results, with one ad per set of search results.

There is no option to turn off ads in the Maps app, but ads are clearly labeled with a blue "Ad" tag. Apple says ads you interact with in the Maps app are not associated with your Apple Account, and data is not shared with third parties.

Apple has been preparing to launch ads all summer, with signs of ads showing up in iOS 26 updates. Apple started allowing businesses to purchase ad slots in August, offering a 15 percent credit toward future ad spend as an incentive.

Ads for home services, bail bonds, and cryptocurrency ATMs are not allowed in the Maps app per Apple's Advertising Services policy. Ads that promote or reference medical services may be allowed, and are evaluated on a case-by-case basis. Maps ads also have to adhere to Apple's other advertising rules.
This article, "Apple Maps Ads Are Here, and There's No Way to Turn Them Off" first appeared on MacRumors.com

Discuss this article in our forums