With more patches per month than at a pirate convention, the bug must be an endangered species. Well, about that
It is the best of times, it is the worst of times – especially if your job is keeping systems patched and up to date. Microsoft has gone from 60-90 Windows security fixes per month last year to a record of 600+ this July. Oracle and Linux are following the same path, and they are very much not alone. The good news is that a lot of bad things are getting fixed very quickly. The bad news is that patches can bring side effects of their own.
There are two mechanisms at work, both driven by the source of and solution to all our woes, AI. The first is that the appropriate LLMs and their humans have got very good at bug hunting. Like demon archaeologists, they’ve started thrashing their way down through the stratified layers of long-established code bases, bringing a huge backlog of previously buried bugs to the surface.
Advertisement
Complicating matters, LLMs are also writing an awful lot of code, some of which is not very good. It is making its way into production for all the old reasons – marketing-led deadline pressure, shape-shifting specs, and Brownian goalposts – until the implacable hostilities of reality spit it back out.
The result is a very interesting dynamic of conflicting pressures that is changing the nature of patches. It’s easy to assume that the current explosion of bug fixes will die down as the code bases are repeatedly refined and purified, and that this time next year we’ll be seeing rather fewer patches than in the pre-AI days, let alone today. It’s a nice thought. Similarly, with the old code in a new state of grace, attention can turn to properly generating and testing the AI-powered stuff, so that it too calms down.
Other factors will work against this. Newer models may find new classes of bugs or start refactoring for efficiency or structural reasons. Not all patches fix bugs, and not all bugs are vulnerabilities. CVEs are easy to count, but aren’t the full story. The pressure to release early won’t go away either; better tools often encourage greater recklessness. Vibe check, anyone? Finally, the bad guys aren’t going away and will be using all the new shiny to keep up their side of the arms race.
This whole system of conflicting pressures in a morphing environment has not been well studied, and the future shape of patching is unclear. One analogy suggests itself, that of stellar evolution.
Advertisement
Astrophysics fans know the score. After a star condenses out of gas and dust, gravity compresses its core until it becomes hot and dense enough for nuclear fusion. Hydrogen nuclei fuse to create helium, releasing energy that pushes outward against the gravity trying to squeeze the core further, and the star shines steadily. When the hydrogen in the core runs low, that balance changes. Depending on the star’s mass, it may begin fusing helium and successively heavier elements before fusion becomes impossible. The possible endings include explosions visible from other galaxies, black holes, neutron stars, cooling relics, and more.
In this analogy, patch generation is fusion pressure, bug generation is gravity, and the nature of bugs and patches evolves as the two interact and the code changes. If any unit of code, no matter how badly written, can contain only so many bugs, then the model tends toward the white dwarf outcome: a remarkably long-lived object that passes the rest of its existence without drama or intervention. It no more needs patching than a pebble does.
It is certainly true that, despite the best efforts of many, code design and implementation are ultra-reliable compared with the days when Windows BSOD’d every other day – and on the hour if you installed drivers – and Big Three PC database company Ashton-Tate’s industry nickname was Crashed and Late. If the object of the industry was to produce pristine versions of, say, Windows 10, then the white dwarf patchless future would be the most plausible.
That is not the industry objective. If a star is big enough, its ending can be a supernova birthing a black hole, a singularity beyond observation where gravity has won. In this case, the battle to write ever-more complex yet bug-free and optimal code is locked in the attempts to find ways to break it, either as part of the production pipeline or in adversarial attacks. If models advance as hyped, iteration times could become so short, and constantly morphing production code so difficult to analyze, that the very model of patching breaks down. The daily build becomes the product, and you get the latest version every time you run it.
That may seem an extreme cosmology, but it’s not so far from what happens every time you fire up a cloud app. You’ve never had to patch Google Docs, but you’ve had features appear and disappear overnight without explanation or warning. This, then, may be the shape of patches to come, a universe where the increasing power of coding and testing models enables new and stranger commercial pressures to modify the software you depend on.
Advertisement
You don’t have to plot that path. Some software has a more steadfast physics. Not for the first time, those who navigate by the constant star of open source may have the safest voyage. ®
Stilo S.R.L. was founded in 1999 by former rally driver Ludovico Fassitelli and his wife, Elena Perini, who built the brand around integrated communications, noise reduction, and motorsport-focused design. In 2016, Texas–based Simpson Performance Products, Inc. acquired Stilo, bringing the Italian specialist under the broader Simpson Performance Group umbrella while keeping its management and operations in Italy. Simpson later took over as Stilo’s official U.S. importer and seller, consolidating distribution and marketing on this side of the Atlantic. In 2021, Simpson and Stilo joined the Holley Performance group, a publicly traded company.
Today, Simpson positions Stilo as its premium auto racing helmet line, alongside other performance safety gear, with Stilo-branded products promoted through Simpson Race Products channels in North America. The brand’s ownership structure matters for buyers because it combines Stilo’s motorsport pedigree with Simpson’s scale in support, homologation, and distribution. For racers, that means easier access to spare helmets, service, and consistent certification coverage, even though the helmets themselves remain distinct in design and positioning from Simpson’s legacy lines.
Advertisement
Where are Stilo helmets made?
Henfaes/Getty Images
Stilo states its helmets are “designed, developed, prototyped, tested and produced in Italy” and emphasizes a “KM 0” approach, meaning sourcing and manufacturing as locally as possible and keeping the entire process in-house. Each helmet is hand‑made by specialists in the Bergamo area, from shell fabrication to assembly, finishing, and labeling, with multiple quality checks along the way. Retailers and Stilo’s own product materials consistently describe the lineup as engineered and manufactured in Italy, and much like the different types of motorcycle helmets, Stilo’s offerings come in a variety of styles like karting and carbon models.
For enthusiasts, that means you’re effectively getting an Italian-built motorsport helmet backed by a U.S.-owned parent company with a strong footprint in racing safety (a key feature to keep in mind when shopping for a racing helmet). Stilo’s “Made in Italy” stance isn’t just a marketing tagline; it’s confirmed in the company’s FAQ and echoed by distributors who highlight Italian construction as a key differentiator in a market where some competitors split design and production across regions.
Having rejected Apple’s initial compromise proposals for App Tracking Transparency in the App Store, Germany says it has made a final agreement that will make the system fair to all developers.
Germany’s federal cartel office, the Bundeskartellamt, has been challenging Apple over its App Tracking Transparency (ATT) feature. This is the feature that requires users to confirm they allow data tracking by developers, and which is seen as favoring Apple’s own apps.
Apple has previously offered a compromise that chiefly consisted of rewording the warning dialogs displayed when a third-party app is first launched. That compromise was rejected but now the regulator says that it has agreed binding commitments with Apple and has concluded its case.
“Apple is allowed to provide for a level of protection for its users that exceeds the minimum legal requirements,” said Andreas Mundt, President of the Bundeskartellamt in a statement. “However, if Apple sets up additional rules within its ecosystem for the use of data, these rules must, under Germany’s special abuse provision for large digital companies, not treat its own offerings better than those of its competitors.”
Advertisement
“This is precisely where our competition concerns arose,” continued Mundt. “Apple will now align the consent requests much more closely and give third-party app providers more freedom to combine the necessary requests in a sensible way.”
What happens next
Apple now has four months to implement the agreed changes to how ATT works. Those changes are binding commitments that the regulator says run for the next seven years, and will be monitored by a trustee.
It’s not clear exactly what Apple has agreed to that is different to its compromise offerings, however. The new rules are again specifically about the wording of dialogs shown to users.
According to the regulator, the specifics of its ruling are that “Apple will align the consent prompts for its own offerings and for third-party apps much more closely. “
Advertisement
That means “removing possibly discouraging symbols and wording in Apple’s predefined requests for third-party providers.” These prompts must always now be “neutral in terms of content, wording and layout.”
The clearest difference is that developers will be allowed to add more detail about their proposed use of data tracking.
Apple does definitely use ATT to discourage users from granting permission for third-party apps to track them. It also does not do that with its own apps, but the company has previously argued that this is because none of its apps even attempt to track users.
At a lower cost, each of Raj’s delivery meals consistently tastes better than the (admittedly not great) Punjabi restaurant down the street from me—less sugary, more rich, with saffron-tinged basmati and chicken that remains juicy and plump, not overcooked. Raj’s Indian gravies offer the most compelling argument for me to order CookUnity, in a city (Portland) where the best Indian food is often a half-hour drive into the suburbs.
But, of course, Raj’s meals are available only to those served by the Seattle hub. Others, near Los Angeles or Atlanta, may have other local chefs who offer similarly compelling arguments. I wouldn’t know.
Local Mileage May Differ
CookUnity Beef Stroganoff with Truffle & Pappardelle from chef John DeLucie
Photograph: Matthew Korfhage
The strength of CookUnity is that each regional hub is basically local. But this is also a potential weakness: Each regional hub is different. Each hub has greater or lesser variety. Each also has a different selection of chefs and meals.
Advertisement
But my good experience with the Seattle hub might translate to what you’ll find in a different kitchen, with different cooks. This may be true even if the same chef’s name and face are on your meal box. Some chefs operate four or five kitchens at different hubs, and different kitchens may have different results, even with the same recipes. And heck, I’ve had different results even from the same chefs. I can only speak to my own good experience and the good experience of my colleague Louryn Strampe, who ordered out of the Chicago hub.
Note that while CookUnity offers a much bigger caloric range than most prepared meal kits, usually the big feasts are heavily carb-loaded: rice, noodles, pasta. Those looking for a focus on proteins and veggies can find them easily, but usually in more moderate portions. Food cost is food cost. Starches are cheap.
But I have reason to believe that CookUnity’s other hubs also offer good cooking—and that has a lot to do with CookUnity’s business model. The overall factor that I think makes CookUnity work is healthy competition.
Each CookUnity chef making prepared meals is, essentially, a competing food business within the same big commissary kitchen. Their name is on each box, and a good number of the chefs have restaurants and reputations to protect. There are many incentives to make each meal interesting and delicious. My experience bears this out.
Amazon has “reintroduced a clause in its user agreement that seeks to prevent shoppers from filing class-action lawsuits against the online retailer,” reports Bloomberg, “inserting a legal buffer between itself and plaintiffs attorneys that it removed five years ago.”
In an email sent to customers on Friday, the company said a new “arbitration agreement and class-action waiver” will require shoppers to resolve disputes outside the courts but said they could still file small claims, cases that typically limit damages to a few thousand dollars… The user-agreement update isn’t necessarily binding in court. Plaintiffs attorneys could still seek class-action lawsuits against Amazon, and it would be up to a judge to determine if the user agreement prevents them from doing so.
Amazon was contacted for an explanation by Bloomberg, and provided a statement saying they continually update their wording “to better serve our customers.”
Amazon said they’d “determined” that “reinstating the arbitration clause will offer customers a fast, cost-effective way to resolve disputes while still giving them the option of going to small claims court.”
The cuts were made even after a preview model went rogue and hacked Hugging Face.
Samuel Bovin/Shutterstock
As part of a restructuring, OpenAI reportedly disbanded its “preparedness” team that assesses the potential for catastrophic risks with its models, The Financial Times reported. The Sam Altman-led company is said to have made the move at the end of last month, despite the fact that several of its models recently went rogue and hacked the AI tool repository, Hugging Face.
Senior staff within separate teams have now been assigned responsibility for different areas of preparedness like bio and cyber, according to the article. OpenAI described the staff cuts as part of a “streamlining process” ahead of its IPO, after Altman asked employees to cut back on “side quests” and focus on its core ChatGPT business.
OpenAI has put some higher-profile side quests on the chopping block of late, recently eliminating its Sora video generation app that became famously associated with AI “slop.” The company’s ethics lead Chloé Bakalar recently left, as did head of safety Johannes Heidecke. Those two departures in particularly have led to concerns that the company is ignoring safety in favor of growth, with one AI analyst comparing OpenAI’s revolving safety door to a Harry Potter curse.
When Apple Intelligence launches in China, it will use a large language model made by Apple, instead of being based on either Google Gemini or the Chinese Qwen.
The Chinese version of Apple Intelligence will utilize Qwen, but according to Reuters, the main LLM will be one trained by Apple. The training was done in partnership with Alibaba, but this is still a departure from Apple predominantly adopting another firm’s models.
In the US and most territories, Apple Intelligence uses Google Gemini technology, although its Foundation Models are still entirely Apple’s. Apple has also offered ChatGPT extensions for Siri, but neither that nor Gemini or Anthropic’s Claude are available in China.
China recently approved Apple Intelligence for use in the country, although it has yet to roll out and there may be further regulatory hurdles. Nonetheless, this new report means that Apple is expected to be the first company to launch a proprietary AI model in China.
Advertisement
Even so, Alibaba’s Qwen model is to be included in some way for Apple devices sold in the country. There will also be technology from Baidu, another local Chinese firm.
All of this is necessary because of how complex China’s laws about AI are. In theory, a US firm could comply with the stringent regulations, but in practice an AI company wanting to enter the Chinese market has to partner with local firms.
There is also the issue of censorship in China. It has previously been reported that Alibaba will censor Apple Intelligence in the country, and that will presumably apply to the use of both Qwen and Apple’s own model.
Apple Intelligence is now expected to be launched in China shortly, although Apple has not announced a schedule. The service did briefly launch in March, apparently accidentally, showing that it is ready.
Advertisement
Similarly, Apple Support recently published a guide to using Qwen on a Mac, but then removed it.
China has always been a major market for Apple, although for many years sales of iPhones declined steeply. That trend has recently been reversed, however, with the iPhone 17 range proving to be even more popular in China than Apple expected.
Slop. The word that people have settled on, and it fits.
LinkedIn posts that say nothing. Investor decks assembled from prompts. Sales outreach that feels personalized but went to thousands of people. Conference presentations that look polished but collapse when someone asks a substantive question.
Nobody in a position of power seems to want to say this clearly, so here it is: we are making it ourselves.
Advertisement
Latest Videos FromTechRadar
Emma McGrattan
People racing to scale content with AI tools are burning through the one thing that makes any of this worth doing, which is the reasonable expectation that when a person puts their name on something, a person thought about it.
Advertisement
When the thinking did not happen, the name is borrowed credibility, and audiences tend to feel the difference.
A denial of service attack on professional attention
Think about what high-volume, low-signal communication actually does to the people on the receiving end. It is a denial of service attack. Done at scale, they do not just waste time. They degrade the channel for everyone trying to use it honestly.
Advertisement
Trust in professional life is not a soft value. It is the mechanism by which work gets done. Slop corrodes that mechanism slowly, quietly reshaping expectations until the default posture toward professional content shifts from curiosity to suspicion.
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
The deck nobody built
Anyone who spent serious time at an event this year noticed it. You sit in the room and thirty minutes in you realize the presenter has not actually field-tested any of what is on the screen. The model made it coherent. Coherence is not the same as true.
Advertisement
The tell is always the same. Someone asks a follow-up question that goes one layer below the framework and the answer drifts. Not because the presenter is unintelligent. Because there was no thinking to retrieve.
This matters because a presentation is a claim: I know something about this, I worked it out, and it is worth your time. An AI-assembled deck without the underlying experience is not a weak presentation. It is a lie. It borrows the authority of hard-won perspective without having won it.
Advertisement
The prototype that is not a product
For most of software’s history, the gap between a prototype and a product was real and substantial. Building a finished product required engineering time, design iteration, infrastructure decisions – all the unglamorous work that separates a demo from production. That gap created useful friction.
AI has compressed that gap in ways that are genuinely exciting and genuinely dangerous. You can now generate a working interface, complete with plausible data and coherent navigation, in a matter of hours. Teams can test market appetite and validate direction before committing to a full build. That is legitimately valuable.
What is not legitimate is presenting that prototype as though the hard questions have already been answered. Can it scale? What happens at the edges? What does it cost to run? These are not details to resolve later. They are the product.
A prototype shown honestly invites conversation. A prototype shown as a finished product forecloses it, and the interest on that debt gets paid in credibility.
Advertisement
Thinking versus the appearance of thinking
The fundamental choice we make every day, often without recognizing it is this: are we using AI to think better or to avoid thinking?
There is a version of AI-assisted work where the human is still the author. The thinking happened. The judgment was applied. Then there is the version that produces slop. The deck assembled via prompts. The prototype shown without the word prototype appearing anywhere. No thinking happened. No perspective was shared. An audience was addressed but did not receive anything.
Getting ahead of the trust crisis means being honest about which version we are practicing and having the discipline to stop calling the second one a strategy.
Advertisement
The bill has started to arrive. Response rates to AI-drafted cold outreach have collapsed. Investors reviewing AI-generated pitch decks are getting quicker at identifying when the strategic narrative was assembled rather than discovered. In product, the reckoning takes the form of churn.
The cruelest part is that the organizations doing the flooding harm not only themselves but everyone trying to communicate honestly in the same space. Individual rationality producing collective ruin, one generated paragraph at a time.
There is a longer consequence that rarely gets named. Frontier AI models were trained on the accumulated output of human knowledge-sharing: forums, papers, articles, the slow sediment of people working things out in public. New questions posted on Stack Overflow are down almost 80% year over year. If the knowledge commons are replaced by AI-generated output, we are not building on a base of human insight. We are iterating on a fixed one.
Slop does not just erode trust in the present. It risks stagnating the knowledge base that the future depends on.
Advertisement
What we owe the people we are trying to reach
As AI drives content toward abundance, credibility becomes the only real differentiator, and it is the scarcest resource in the communication ecosystem. The organizations and individuals who protect it now will hold something that becomes more valuable as the surrounding environment degrades.
Worth the time to actually write the thing, work through the strategy, build the product before calling it one. Worth asking whether what you are about to put in front of people reflects your thinking or just a model’s prediction of your thinking. And sometimes it is worth saying nothing rather than saying the algorithmically optimal version of nothing.
Trust has always been the currency of professional influence. Right now, a lot of people are flooding the market with counterfeits. The people on the receiving end know. They may not say so in the meeting, or click unsubscribe, or walk out of the session. But something shifts. The next email gets opened a little more slowly. The next deck gets a little less benefit of the doubt. The next demo gets a harder question in the room.
Advertisement
Credibility does not collapse all at once. It drains, quietly, one hollow interaction at a time. And by the time you notice the account is empty, the withdrawals have been happening for a long time.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
Alongside the new Pixel 11 series, Google has confirmed that the Fitbit Air will launch in India in October 2026. The wearable stands out as the company’s first fitness tracker with a screen-free design. It was recently launched in select global markets and is now headed to India. Google announced the India availability during its Made by Google event briefing. The Fitbit Air was also spotted on the BIS website and Amazon India earlier.
Google Fitbit Air Specifications and Features
Fitbit Air is a product from Google, designed without a screen to minimize distractions during use. The fitness band does not display any notifications or fitness information. It relies on vibration technology and a little red light for giving alerts. However, one can view all their health data via the Google Health Application. The app also provides activity reports and coaching features.
For health tracking, the Fitbit Air offers features such as continuous heart rate and heart rhythm monitoring. It also includes AFib alerts and SpO2 for blood oxygen level tracking. In addition, the wearable measures resting heart rate and heart rate variability. The device monitors sleep stages and sleep duration to analyze sleep behavior. Moreover, it detects automatic workouts and supports manual exercise logging, as well as Google Health Coach workouts. The battery life is seven days after a single charge. Just five minutes of charge gives an entire day of battery life. The tracker has a polycarbonate body, 50-meter water resistance, and also supports interchangeable bands.
Price and Availability in India
Google has confirmed that the Fitbit Air will arrive in India this October. However, the company has not announced how much the tracker will cost. It carries a $99.99 price tag in the US, which is around Rs. 9,500. The Indian price could differ from the US price, ranging from Rs 17,380 to Rs 18,490.
Google has also kept other availability details under wraps for now. The company has not confirmed the exact launch date, color options, or sales channels. It showcased the Fitbit Air with a pink Performance Loop band. However, this colorway is not confirmed for the Indian market. The company should reveal more information closer to the launch.
High-end audio has always had a complicated relationship with diminishing returns. Spend twice as much and you do not get twice the performance, and there are plenty of products that arrive with heroic price tags, beautiful casework, and very little evidence that your money went anywhere near the sound. Every so often, however, something lands in the listening room that makes the equation considerably easier to defend. You hear where the money went. You can feel where the money went. And, in the case of ATC, you may even come away thinking they gave you rather more than you paid for. How novel.
This review is Part II of a three-part ATC trilogy. Part I featured the larger ATC SCM50ASL Active Loudspeakers, which set the bar extremely high, while the final chapter will involve ATC’s $99,999 EL50 Anniversary Active Loudspeakers, scheduled to arrive here in roughly a month. Having already heard the EL50 at AXPONA 2026, I have a reasonably good idea what is coming, although I will still need to explain to my children why two very large British loudspeakers worth more than any car I have ever owned are occupying valuable real estate. My floors have not been consulted.
Sitting between those two experiences are the considerably more compact ATC SCM20ASL Active Loudspeakers, and they raise a much larger question about where high-end audio is headed. Active loudspeakers are becoming increasingly difficult for audiophiles to ignore, and I am not talking about wireless speakers with streaming platforms, apps, DACs, HDMI inputs, and enough DSP to land a small aircraft. The SCM20ASL are something very different. Their analog signal is divided at line level by an active crossover before being sent to dedicated Class A/B amplifiers for the tweeter and mid/bass driver.
There is no external power amplifier, no passive crossover between amplifier and driver, and no need for loudspeaker cables. You still select the preamplifier, DAC, streamer, turntable, phono stage, CD player, or whatever else you want upstream.
Advertisement
For a hobby that has spent decades convincing itself that more boxes, larger power supplies, thicker cables, and increasingly elaborate amplifier matching are necessary steps on the road to enlightenment, that represents a rather significant change in thinking. An active system can eliminate one or more expensive components, remove the amplifier/loudspeaker matching exercise entirely, and allow the manufacturer to design the crossover, amplification, and drivers as one integrated system.
You still need AC power and interconnects running to each loudspeaker, so nobody has repealed the laws of cable management, but the equipment rack can become considerably less crowded.
And none of this appeared last Thursday. ATC, PMC, Amphion, and Genelec have spent decades moving between professional studios and domestic listening environments, giving their engineers the benefit of feedback from people who actually make records as well as those of us who obsessively listen to them.
So at $10,999 per pair, are the ATC SCM20ASL another expensive lesson in diminishing returns, or one of those increasingly rare high-end products where you can actually hear where every damn dollar went?
Advertisement
Related Reviews:
Build Quality: Bring a Friend and Save Your Back
Calling the ATC SCM20ASL a stand-mounted loudspeaker is technically correct in the same way that calling a Bentley a means of transportation is technically correct. At 17 11/16 x 9 13/16 x 16 3/16 inches (HxWxD), the SCM20ASL does not consume a huge amount of floor space, but each loudspeaker weighs 53.6 pounds before you add a pair of stands.
Treat them like floorstanders. Very heavy floorstanders.
I learned this at a particularly inconvenient point during the review. A severe kidney stone attack landed me in the hospital, and the best part of that particular biltong-and-droëwors sandwich was that the treatment was followed by shingles and a severe rash that lasted for weeks. FUN. The upside, if one can call it that, was a lot of enforced time sitting in one place, listening to music and writing. The downside was having a pair of 53.6-pound active loudspeakers in the room that occasionally needed to be repositioned.
Advertisement. Scroll to continue reading.
Ask a friend to help you set these up. Preferably one you like enough to warn beforehand.
Advertisement
The SCM20ASL cabinets are beautiful in that very British, purpose-driven ATC way. Nobody in Gloucestershire appears to have been asked to make them look like a spaceship or an expensive Italian handbag. The sealed 20-liter enclosure is constructed to be heavy and inert, using internal bracing, damping, and bitumastic panels at critical points to suppress unwanted cabinet resonance.
Knock on the side and you quickly understand that ATC was not looking for ways to save money on MDF or shipping weight. These things are tanks.
And not some cheap Soviet one that a Merkava could drive over on its way to lunch in Kiryat Shmona.
The quality of the cabinetry and veneer work is excellent. Standard finishes include Cherry, Black Ash, Walnut, and Oak real-wood veneers, along with Satin Black and Satin White. Premium options include Rosewood, European Crown Cut Walnut, Burr Magnolia, and Pippy Oak, with various high-gloss finishes, Piano Black, Piano White, and additional special-order options also available.
Advertisement
The removable black cloth grilles are another matter.
They do exactly what they are supposed to do: protect two very expensive drivers from fingers, pets, children, and whatever else might wander too close. They also look like they were designed sometime around 1983. My old Celestion Ditton 33 MKII loudspeakers had something similar, so perhaps there is some historical British precedent here.
I left them off.
The SCM20ASL looks considerably better with its 6-inch Super Linear mid/bass driver and recessed 1-inch S-Spec soft-dome tweeter exposed. You paid for the engineering, so you might as well be able to see it. Covering these drivers feels a little like installing a transparent hood on a muscle car and then throwing a blanket over it.
Advertisement
Amplification and Connectivity
Spin the SCM20ASL around and any illusion that this is merely a very substantial bookshelf speaker disappears.
Most of the rear panel is occupied by ATC’s 250-watt amplifier pack, and it looks like a proper power amplifier because that is precisely what it is. A large, heavy-duty heatsink dominates the rear panel, while substantial handles make it considerably easier to maneuver 53 pounds of British engineering onto a stand without introducing it to the hardwood floor.
The handles are not decorative.
Advertisement. Scroll to continue reading.
Inside are two discrete Class A/B MOSFET amplifiers, with 200 watts dedicated to the mid/bass driver and 50 watts to the tweeter. The amplifiers are convection cooled, hence the formidable heatsinking, and ATC integrates them with an active crossover ahead of the amplification stages rather than sending a full-range amplified signal through a conventional passive crossover.
Advertisement
That distinction matters.
The SCM20ASL is not a passive loudspeaker with an inexpensive amplifier module bolted onto the back so someone can call it “powered.” The drivers, active crossover, amplification, and cabinet were engineered as a complete system.
Connectivity is equally purposeful. Each loudspeaker provides one balanced female XLR analog input and requires its own AC connection. There are no RCA inputs, USB ports, HDMI sockets, streaming modules, Bluetooth radios, or network connections. These are active loudspeakers, not wireless all-in-one systems.
The rear panel also provides controls for input sensitivity and a low-frequency shelf. Input sensitivity can be adjusted from 1Vrms to 4Vrms for full output, while the bass control provides adjustment from -2dB to +3dB in 1dB increments, giving listeners some flexibility when dealing with room boundaries and placement.
Protection circuitry monitors amplifier DC offset and overheating, with a rear-panel LED providing fault indication.
Advertisement
It all feels engineered for decades of use rather than a product cycle.
ATC SCM20ASL Drivers: Two Is Plenty
There are only two drivers staring back at you, but ATC apparently decided that both of them had better be rather good.
The 6-inch SB75-150SL Super Linear mid/bass driver handles everything up to the 2.2kHz crossover point and uses an unusually large 3-inch voice coil with ATC’s underhung motor architecture. Its Super Linear Magnetic System is designed to reduce magnetic distortion in the motor structure, particularly through the bass and midrange where so much of the musical information lives. The driver also offers substantial linear excursion for its size, which becomes rather important once you start asking a compact sealed loudspeaker to behave like something considerably larger.
Above 2.2kHz, ATC’s 1-inch SH25-76S S-Spec soft-dome tweeter takes over. It is also designed and manufactured in-house and uses ATC’s dual-suspension architecture to improve stability and control of the dome and voice coil. The tweeter does not use ferrofluid and is designed for high output, low distortion, and long-term reliability rather than simply chasing an impressive upper-frequency number.
Both drivers are proprietary ATC designs, and that matters. The SCM20ASL is not a cabinet assembled around a pair of off-the-shelf transducers with an amplifier attached to the back. The drivers, amplification, active crossover, and sealed enclosure were engineered as a complete system.
Advertisement
And if you are looking at two drivers in a cabinet this size and wondering whether they can provide enough scale and low-frequency authority for a 16 x 13 x 9-foot or even 20 x 13 x 9-foot room, prepare to put the subwoofer back in the box.
Advertisement. Scroll to continue reading.
These things have considerably bigger lungs than their dimensions suggest.
Setup: 54 Pounds of British Engineering Needs a Proper Perch
One slightly odd omission at this price is that ATC does not offer a dedicated factory stand for the SCM20ASL. ATC recommends using a heavy, rigid, non-resonant stand with the acoustic center at or just above ear level, but leaves the actual hardware choice to you. There are purpose-built third-party options, including the Custom Design OM44 and SQ 404, both offered specifically for the SCM20/SCM20ASL.
Advertisement
None of the stands I had on hand from Gig Harbor Audio, Q Acoustics, or Wharfedale were a good match. The height was workable, but the top plates were too small for a 53.6-pound loudspeaker this deep, and I had zero interest in testing their stability with Tyrion the Westie tearing around the house.
My BDI credenza proved to be the better option because I could pull the speakers farther from the wall, with the tweeters sitting slightly above ear level from the sofa. The second credenza placed the tweeters almost perfectly at ear level, but left the rear of the cabinets only 8 inches from the wall.
The sealed design made that placement less problematic than it would be with many rear-ported speakers, but the BDI setup ultimately gave me more flexibility and better results.
Advertisement
ATC CA2 Preamplifier ($3,749)
Listening
Because each SCM20ASL accepts its analog input through XLR, you need some means of controlling volume before the signal reaches the loudspeakers. That can be a DAC or network player with variable output, or a conventional preamplifier feeding the ATCs through XLR. A fully balanced signal path is not mandatory, although it certainly opens up more options with this type of loudspeaker.
ATC makes two very good analog preamplifiers of its own. The SCA2 is the considerably more expensive flagship and provides genuinely balanced XLR inputs and outputs, while the less expensive CA2, which ATC supplied for this review, uses XLR output connectors but is actually an unbalanced design internally. The CA2 currently sells for $3,749, versus $12,999 for the SCA2, so there is quite a bit of daylight between them. You can obviously use your own preamplifier, and I suspect a lot of SCM20ASL owners will already have something rather good sitting in their rack.
They should.
One of the defining characteristics of the SCM20ASL is just how revealing it is of everything placed ahead of it. That does not mean you need to start burning $100 bills in the driveway before listening, but these loudspeakers are sufficiently transparent that better sources, DACs, phono stages, and recordings produce very obvious dividends. Based on other systems in which I have heard the SCM20ASL and larger ATC active models, I am not entirely convinced that I extracted even 90 percent of what these loudspeakers are capable of with the equipment available to me.
Advertisement
My analog front end consisted primarily of my Thorens TD-160 Super with an Ortofon 2M Black, feeding an E.A.T. phono preamplifier. Digital sources included the Bluesound NODE Icon Network Player with my Holo Audio Spring 3 DAC, along with the Cambridge Audio CXN100 Network Player. I even inserted the WiiM Ultra into the system because inexpensive digital front ends deserve their day in court as well. Audiolab and Marantz CD players and transports handled physical media.
Interconnects, power cords, and digital cables came from Cardas, Audio Art Cable, QED, Analysis Plus, and Clarus Audio, and I swapped enough of them during the review period to confirm something more important than which cable cost what: the SCM20ASL does not obscure changes upstream.
Advertisement. Scroll to continue reading.
And that is where reviewing them becomes dangerous. The better the source became, the more the ATCs revealed, which raises the uncomfortable possibility that the $10,999 loudspeakers are not the component in your system you should be worried about.
Advertisement
No Warm-Up Required
One of my few nitpicks with the larger SCM50ASL was that they needed some volume before everything really started to click. At lower listening levels, some of their clarity, dynamic presence, and overall magic receded; once I became more liberal with the volume control, they were magnificent. The SCM20ASL did not behave that way. Right from the get-go, with Nick Cave, Jonatan Alvarado, and Jason Isbell, these smaller ATCs sounded fully engaged, even when I wasn’t trying to shake the furniture.
Something else became obvious very quickly: bad recordings will not be handled with tender hands. If a recording sucks, the SCM20ASL will let you know exactly how and where. Fortunately, the material I started with was well recorded and provided three very different male voices with which to judge tonal accuracy, texture, image stability, and the speaker’s ability to make a human voice sound like an actual human being.
Nick Cave’s growl, edge, and natural weight were exceptionally well served by the ATCs. There was clarity and inner detail for days, but none of it sounded clinical or dissected, which might surprise anyone expecting a loudspeaker with deep professional-audio roots to behave like a laboratory instrument. Cave had texture, chest, and physical presence, with enough impact to make his voice feel as though it had entered the room and hit me with a car. Nothing enormous. A Fiat 500 or Mini Cooper. But you’re definitely exchanging insurance information afterward.
“Avalanche” has always hit me differently depending on where my head was at when I listened to it, and the story behind the song connects with me on a very personal level. Through the SCM20ASL, the piano was reproduced with a combination of delicacy, weight, and percussive power that is extremely difficult to get right. Piano can expose a loudspeaker very quickly because the attack, harmonic structure, decay, and sheer physical weight of the instrument all have to work together. The ATCs created a rock-solid central image and allowed individual notes to carry both their initial impact and the quieter information that followed.
Advertisement
The soundstage in my system was not especially deep, although I am inclined to blame some of that on my placement limitations rather than the loudspeaker. Image specificity, however, was superb.
Argentine tenor Jonatan Alvarado provided a very different test. His lighter, almost ethereal vocal presentation was positioned noticeably farther back in the soundstage, which is precisely where I expected it to be. I have listened to the same recording through more than 40 pairs of loudspeakers, and only a handful have recreated that spatial relationship convincingly. The Bowers & Wilkins 703 S3, which I reviewed in 2023, was one of the best in that regard; the SCM20ASL belongs in the same very short conversation. The image was stable, appropriately distant, and frighteningly human.
Shivers down the spine.
Jason Isbell brought an entirely different kind of intensity. He remains one of the more underrated American songwriters of his generation, with a style that moves comfortably through country, rock, folk, and Southern roots music, but it is the combination of his guitar work and forceful vocal delivery that I find so compelling. Through the ATCs, that power was not created by pushing his voice artificially forward or sharpening its edges. The SCM20ASL preserved the texture, scale, and emotional force of the performance without turning it into a demonstration of how much detail the speaker could retrieve.
Advertisement
That might be the most important point.
The SCM20ASL are exceptionally revealing loudspeakers, but revealing and clinical are not the same thing. These speakers can dissect a poor recording without mercy and then turn around and make a great vocal recording feel unsettlingly human.
That is a much harder trick.
Advertisement. Scroll to continue reading.
Advertisement
Put the Subwoofer Back in the Box
My primary listening room can be overloaded by loudspeakers with too much low-end energy, especially when they are pushed too close to the corners. Because the SCM20ASL were not positioned where most of my review speakers normally reside, I wasn’t entirely sure what to expect.
Unless your listening room is a cavern or some enormous open-concept loft, I suspect most listeners will find the SCM20ASL deliver more than enough bass on their own. The low end is tight, quick as hell, full, and genuinely impactful, without sounding thick or artificially enhanced.
I’ve been on something of a Rush kick lately because of the Fifty Something Tour, and “Xanadu” seemed like an entirely sensible way to determine how hard these little British tanks could be pushed. At what my children would describe as “WTF is wrong with you, Daddy?” levels, the SCM20ASL barely flinched. Alex Lifeson, Geddy Lee, and the late-Neil Peart remained firmly separated across the soundstage, with the scale, speed, and impact of the performance intact. “Xanadu,” from A Farewell to Kings, remains one of Rush’s great showcases for the interplay between all three musicians.
I was air drumming like the nerdy teenager who first saw Rush at Maple Leaf Gardens in the early 1980s.
Metallica, AC/DC, and The Police only reinforced the point. The SCM20ASL could play hard and extremely loud without turning everything into a congested wall of sound, and bass lines remained easy to follow even when guitars and percussion were going full throttle. Unless your objective is to have your significant other put something lethal in the biltong that just arrived from South Carolina, there is a lot more output here than the cabinet dimensions suggest.
Advertisement
Switching to Peter Gabriel’s “Sledgehammer,” Talking Heads’ “Born Under Punches,” Boards of Canada’s “Roygbiv,” Aphex Twin’s “Xtal,” The Orb’s “Little Fluffy Clouds,” and Kraftwerk’s “The Robots” revealed another strength. The ATCs could unravel dense layers of synthesizers, electronic percussion, bass, effects, and vocals with exceptional specificity while still presenting them as a coherent piece of music.
The SCM20ASL can pull a recording apart with almost forensic precision, but it does not leave the pieces scattered across the floor. Synths had weight and texture, electronic percussion had real snap, and the top end remained highly revealing without becoming hard unless the recording itself was dreck.
Timing, Texture, and No Place to Hide
I do not live in a house that particularly likes jazz. My kids are far more interested in MF DOOM, Tyler, the Creator, and about 30 other artists I couldn’t identify in a police lineup even if they were wearing name tags. And I should know. I’ve been in a few.
Jazz is therefore a late-night or very early-morning thing for me. I’m up at 5 a.m. every friggin’ day, which gives me a narrow window before the rest of the house starts moving. I make some Rooibos tea; I’ve stopped dipping the biltong, leave the lights off, and wait for the sunrise over the ocean five blocks away to slowly illuminate the room.
Advertisement
Then I close my eyes and think.
About the good things I’ve done. The bad things. The things I would change if the universe offered refunds. And, more often than I probably should, about one very specific person. Apparently this is how the universe has chosen to punish me. Some of it is probably deserved.
Miles Davis, Lee Morgan, McCoy Tyner, Donald Byrd, and Eric Dolphy have become the soundtrack for those hours, and this is where the SCM20ASL did something that very few loudspeakers manage. It could reproduce the intimacy, tension, anger, remorse, regret, and longing inside these recordings without turning any of it into an audiophile demonstration.
Advertisement. Scroll to continue reading.
Advertisement
The ATCs let you hear frighteningly deep into a performance, but what matters is that they never forget there were actual people behind those notes. Imperfect people. Complicated people. Musicians trying to communicate something they perhaps could not say any other way. You hear the breath behind a horn, the weight of a piano key, the tension in a bass line, the decay of a cymbal into the room — but none of it feels clinical or sanitized.
There is something almost visual about it. A dim room. A stool. A piano. An ashtray somewhere nearby. The remains of a long night hanging in the air. Not because the ATC manufactures atmosphere, but because it preserves enough of what was captured to let your brain fill in the rest.
Very few loudspeakers I have heard can do that this convincingly. Maybe one percent.
The SCM20ASL can sound enormous when the music demands it, but its real gift may be how small and intimate it can become when there is nowhere for the emotion to hide.
Advertisement
The ATC understands that the notes are only half the story.
The Bottom Line
The ATC SCM20ASL are among the finest active loudspeakers I have heard because they combine studio-grade transparency with the kind of texture, weight, intimacy, and emotional connection that far too many “revealing” speakers manage to lose along the way. Their active architecture removes the amplifier-matching game entirely, the build quality is exceptional, and their ability to scale from late-night jazz to Rush at antisocial volume levels is genuinely impressive.
What are they missing? There is no wireless connectivity, streaming platform, digital input, room correction, or RCA connection, and ATC really should offer a dedicated stand for a loudspeaker this heavy and expensive. They also expose poor recordings and mediocre sources without much mercy.
At $10,999 per pair, the price is substantial, but context matters. Passive loudspeakers capable of this level of resolution, dynamics, and tonal accuracy are not inexpensive, and adding a truly comparable power amplifier can make the final bill considerably uglier.
The SCM20ASL are not cheap.
Advertisement
They are, however, one of those rare high-end products where it is very easy to hear where the money went.
Pros:
Exceptional transparency without sounding clinical
Superb dynamics, speed, and low-frequency control
Excellent performance at both low and high listening levels
Proprietary ATC drivers and dedicated Class A/B amplification
Outstanding build quality and cabinet construction
Reveals source and recording quality with remarkable clarity
Can deliver serious scale without requiring a subwoofer in most rooms
Removes the cost and guesswork of amplifier matching
Cons:
$10,999 per pair is still a significant investment
No wireless connectivity, streaming, DAC, room correction, or digital inputs
XLR-only analog input limits some system configurations
Poor recordings and mediocre sources get very little forgiveness
Every time I switch from any Android phone to a Pixel, there’s one small gesture missing that always annoys me. From a modern Galaxy phone or even a device from Xiaomi, I can double-tap an empty part of the home screen and immediately turn the display off. Samsung has offered a double-tap to turn off the screen since Android 11. On a Pixel, I still have to reach for the power button.
To be fair, Google isn’t the only offender, and Motorola’s stock Android phones, like the Razr Fold, also skip such a basic feature. But this might be changing soon. Android Authority has discovered updated code strings inside Android 17 QPR2 Beta 3 pointing to Google’s long-in-development double-tap-to-sleep feature. Better still, Google appears to have expanded it beyond the lock screen and onto the home screen, which is exactly where I want it.
Nadeem Sarwar / Digital Trends
I need this to ship ASAP
Double-tap an empty area on either the home screen or lock screen, and the display turns off. Android Authority also found strings referring to a dedicated settings page where users could enable or disable the gesture. Google had previously been working on a narrower implementation, which was initially uncovered in Android 16 QPR1 Beta 1 last year.
Nadeem Sarwar / Digital Trends
Back then, its description only mentioned double-tapping an empty area of the lock screen. More than a year later, the updated string now explicitly mentions the home screen as well. There’s even underlying support for this concept in Android’s framework. AOSP contains a “config_supportDoubleTapSleep” setting describing double-tap gestures on non-action areas of the lock screen and launcher workspace, but it is disabled by default in the referenced configuration.
Almost every other brand made me want it
I know double-tapping a screen to turn it off sounds painfully minor. After using phones that support it, I miss it almost immediately when I go back to a Pixel. It lets me put my phone down, tap the screen twice, and walk away. This is a lot more convenient than always having to press the power button. Larger phones get the most benefit; picture a model like the Pixel 11 Pro XL or Pixel 11 Pro Fold, since you don’t have to reposition your hand just to hit the power key.
Vikhyaat Vivek / Digital Trends
Google is arriving very late to this particular party. And yet, you still can’t actually enable the feature on a Pixel–at least not yet. This was discovered only through an APK teardown, and work-in-progress strings are never a guarantee that something will reach the public release. Google’s official Android 17 QPR2 Beta 3 release notes don’t announce the gesture either.
So I’m hoping that changes quickly. Android 17 already gives Google plenty of opportunities to make the Pixels smarter. Though sometimes the real quality-of-life upgrade is just a simple tap or two. With the latest Pixel 11 series devices having models that cost north of $1000, missing something so basic is pretty frustrating.
You must be logged in to post a comment Login