Technology

Your tablet could double as a second monitor. Here's how Samsung Second Screen and Apple Sidecar work, and when a dedicated display is worth buying.

The post Before You Buy a Second Monitor, See if Your Tablet Can Do the Job appeared first on TechRepublic.

Singapore’s Swap Mobile offers phone upgrades every six months, with flexible handset options and a new 24-month contract for each eligible switch.

The post Singapore’s Swap Mobile Lets You Change Phones Every Six Months — Here’s How It Works appeared first on TechRepublic.

Google's Gemini 4 Carbon brings advanced cybersecurity capabilities to selected defenders first. Here's who's eligible and what comes next.

The post Google Unveils Gemini 4 Carbon With Advanced Cybersecurity Skills — Who Gets Access First? appeared first on TechRepublic.

Realme has revealed when 49 phones are scheduled to receive ColorOS 17, with the first Android 17 updates starting October 29 and more arriving through 2027.

The post Realme ColorOS 17 Rollout Starts: Full Update Schedule for 49 Phones appeared first on TechRepublic.

The M6 MacBook Pro costs $900 more than the M5 MacBook Air. Compare the display, ports, storage, and performance claims before choosing.

The post M6 MacBook Pro vs M5 MacBook Air: What Does the Extra $900 Buy? appeared first on TechRepublic.

• US Air Force wants AI to detect mistakes during aerial refuelling
• Tanker cameras would supply footage for automated analysis of refuelling operations
• AI would monitor aircraft speed, alignment, positioning, and approach velocity

The US Air Force is exploring using AI to identify mistakes during aerial refuelling operations and improve how crews receive performance feedback.

The proposed system would analyze footage captured by tanker aircraft cameras, identifying flight manoeuvres and operational errors that conventional assessments might overlook.

Military officials want the technology to provide detailed evaluations of refuelling exercises, reducing dependence on human judgement during one of aviation's most demanding procedures.

AI system will monitor aircraft movements and refuelling accuracy

The initiative focuses on developing software capable of examining interactions between tanker aircraft and approaching receivers throughout the complicated fuel transfer process.

Using computer vision and machine learning, the technology would examine aircraft movements and determine whether crews maintained appropriate separation and alignment.

It would also monitor the refuelling boom, comparing its movements against aircraft trajectories to identify potentially dangerous deviations during fuel delivery.

Measurements would include aircraft speed, relative positioning, approach velocity and alignment, allowing instructors to examine specific manoeuvres rather than relying on general observations.

The proposed software would automatically identify departures from expected flight conditions and generate performance information for instructors reviewing completed refuelling missions.

Currently, assessments depend substantially on instructor observations, crew evaluations and manually examining recorded footage, according to the military's research solicitation.

These methods can produce inconsistent judgements, particularly when instructors interpret the same manoeuvre differently or overlook brief errors during complicated operations.

Therefore, the software’s assessment will include graphical overlays, colour-coded performance maps and chronological displays identifying significant events during individual refuelling exercises.

The Air Force consequently wants automated measurements that could provide more consistent evaluations while helping personnel understand precisely where mistakes occurred.

The project falls under the Small Business Innovation Research programme, which invites companies to develop technological solutions addressing specific military requirements.

Interested developers have until November 25 2026 to submit applications outlining how their proposed systems could satisfy the service's operational and training requirements.

KC-46 tanker technology and future training applications

The initial development effort concentrates on the KC-46 Pegasus, a refuelling aircraft equipped with cameras supporting remotely controlled fuel transfer operations.

Its remote vision equipment allows boom operators to guide the refuelling apparatus using camera feeds and 3D displays instead of direct observation.

However, the aircraft's existing viewing technology has experienced persistent technical difficulties, although military officials expect most outstanding problems to be resolved soon.

Developers are expected to consider interactive review environments that allow pilots and instructors to examine manoeuvres from different perspectives.

The technology could eventually connect with flight simulators, virtual reality equipment and augmented reality training platforms already used for aviation instruction.

Military planners want an operational prototype suitable for integration with existing training arrangements for at least one category of receiving aircraft.

The solicitation also identifies potential applications beyond refuelling, including automated assessments of other aviation activities throughout the Department of Defense

Via Defense News

Elvis Presley has been dead for nearly 50 years, but apparently even the 'King of Rock and Roll' can't escape the AI industry's appetite for training data. The owners of 274 Elvis songs made famous by Presley have sued Anthropic, accusing the company behind Claude of using his music without permission to build its AI models, and are seeking tens of millions of dollars in damages.

AI companies should have to take copyright seriously too, even if that makes their business models more expensive.

Anthropic disputes the allegations and says it intends to defend the case. AI companies have long argued that training models on copyrighted works can qualify as fair use, an issue courts are still working through. There's an important legal difference between learning patterns from existing material and reproducing protected expression, but that distinction has become a remarkably convenient place for the AI industry to hide its enormous dependence on other people's creativity.

It's hard to sympathize with the companies on the receiving end of these sorts of lawsuits, even when the plaintiffs are not without resources of their own. Unfortunately, these disputes are becoming all too normal. AI-powered music is becoming more widespread, if not necessarily popular, and we've reached a point where discovering that an AI company allegedly used hundreds of copyrighted songs without permission barely qualifies as surprising news.

The only unusual part is that this time the person whose music is at the center of the dispute happens to be Elvis Presley.

A little less uncompensated song use

The lawsuit makes an accusation that goes beyond simply using copyrighted material for training. According to the complaint, Claude could generate material that closely resembled existing Elvis songs, even after Anthropic introduced safeguards intended to prevent infringement. Requests to recreate songs for modern audiences allegedly produced versions retaining recognizable elements of the originals.

Claude is an impressive assistant, but its ability to answer questions doesn't suddenly make songwriters' copyright interests irrelevant.

An AI model learning about music isn't automatically infringing copyright, and copyright law doesn't give anyone ownership of an entire musical style. But if a commercial AI system reproduces protected lyrics or creates unauthorized adaptations, it's hard to see why the company behind it should get a special exemption simply because the copying happens through complicated software.

AI companies have been trying out various arguments in court, but a lot of it feels like an attempt to claim that AI development is so important that the usual rules should somehow bend around it. Training requires enormous amounts of information, and licensing everything would be expensive or inconvenient, but that's a standard headache for companies whose products rely on copyrighted material.

(Image credit: Warner Bros)

TV shows and movies license music, whether an original track or just the right to record a new version for their project. No company is allowed to simply grab hold of copyrighted material simply because clearing it would slow down development. AI companies should have to take copyright seriously too, even if that makes their business models more expensive.

It's also particularly galling that songs people spent years creating and popularizing can become raw material for commercial technology without the rights holders having any meaningful say. Access to a work and permission to exploit it are separate questions, regardless of how sophisticated the software becomes. Ignoring those rules is why artists who passed away decades ago seem to produce new music whose earnings don't go to their estate.

AI music turns the volume up

Elvis is hardly the first musician to become part of this fight. Major music publishers have already sued Anthropic over allegations involving copyrighted lyrics, while AI music generator Suno has faced legal challenges from record labels. The Elvis lawsuit stands out because it focuses on the catalog associated with one of the most recognizable performers in history.

That fame might help draw attention to a problem that has affected considerably less powerful creators for years. Independent artists may struggle to establish whether their work was used to train a model. If someone with Elvis Presley's commercial legacy needs a federal lawsuit to challenge alleged unauthorized copying, ordinary musicians' prospects look bleak.

Of course, the companies controlling Elvis's catalog aren't struggling artists themselves, but that doesn't make the underlying complaint any less important. The principle shouldn't depend on whether the copyright owner is wealthy or famous.

(Image credit: Getty Images / Mike Kemp)

I'm also not convinced that every AI company needs unrestricted access to practically everything ever written or recorded to produce useful technology. Claude is an impressive assistant, but its ability to answer questions doesn't suddenly make songwriters' copyright interests irrelevant. If AI developers want to build profitable products using creative works, they should be prepared to explain where that material came from and why they were entitled to use it.

The courts may decide in Anthropic's favor due to fair use, but even that ruling wouldn't automatically mean AI companies have permission to go wild on every song ever recorded. The best path forward won't be as simple as making it always allowed or illegal to use copyrighted material for AI training, but the companies building these systems shouldn't rely on their ambitions as a good enough reason to ignore rights holders.

The AI industry has become very good at explaining how much it needs human creativity, but it's time it became equally good at respecting the people responsible for it.

With Google now injecting “sponsored” links directly into AI follow-up queries, traffic to online publishers will suffer even more.

Celebrate the life of Patrick Stewart through his iconic roles in TV and movies.

• Nvidia reveals plans to invest $3 billion in Meta’s share offering
• Meta could raise $95 billion to finance its expanding AI infrastructure
• Nvidia hardware will support Meta’s next generation of artificial intelligence agents

Nvidia has revealed plans to invest $3 billion in Meta’s share offering while both companies increase cooperation on artificial intelligence systems designed to operate as personal agents.

The agreement will give Meta access to additional Nvidia computing technology as Facebook’s parent company develops its Muse models and associated artificial intelligence services.

It also includes discussions about agent supervision technologies, although the companies have not disclosed a timetable for implementing every proposed collaboration.

Nvidia hardware will power Meta’s next generation of AI agents

Meta expects to install more Nvidia equipment, including its forthcoming Vera Rubin platform and Vera central processors, to support training and execution of advanced agent systems.

The companies also intend to improve performance throughout the computing process, with particular attention to processing efficiency, system throughput and infrastructure operating expenses.

Networking equipment will form another part of Meta’s planned deployment, including Nvidia Spectrum-X technology and BlueField-4 components for handling communication between computing resources.

The companies are considering Nvidia’s OpenShell architecture for safer agent deployment, alongside Sentry technology intended to track agent activity and support data processing.

“AI agents represent a new computing paradigm — software that understands our goals, reasons, uses tools and gets work done on our behalf,” said Jensen Huang, founder and CEO of Nvidia.

“This changes how we use computers, from operating applications to working with intelligent agents.”

Meta claims its Muse systems have achieved strong results in benchmark testing, while consumer downloads have climbed since their initial release.

“Muse is how we bring personal superintelligence to everyone. We’re excited to keep working with Nvidia to deliver the best personal agents and frontier models for billions of people,” said Mark Zuckerberg, founder and CEO of Meta.

However, the companies have not supplied specific download totals or independent comparisons that would establish how widely these models outperform competing products.

Meta’s multibillion-dollar fundraising supports its AI expansion

Nvidia’s investment comes as Meta prepares to raise substantial capital through share sales intended to finance its expanding artificial intelligence infrastructure.

Meta initially outlined a fundraising programme worth approximately $35 billion, comprising $22.5 billion in ordinary Class A shares and $12.5 billion in convertible depositary securities.

The company subsequently increased its ordinary share sale to $25 billion, with each Class A share offered at $625 and depositary securities priced at $50 per share.

Its broader fundraising plans could generate up to $95 billion through successive offerings, though the full amount is not guaranteed.

Nvidia’s proposed $3 billion contribution would make it a participant in Meta’s financing, linking its investment directly to a major customer’s infrastructure spending.

The companies have not disclosed how much additional Nvidia equipment Meta will purchase under the expanded partnership or when all planned deployments will occur.

The personal information of more than 41,000 Department of Homeland Security employees was exposed by a hacker group.

• Getac's B360X and B360X Pro are bringing Intel Panther Lake chips, 14-inch 1,400-nit screens, and Wi-Fi 7 to its rugged B360 line
• The 180 TOPS headline only applies to the B360X with a Core Ultra X7, a chip Getac doesn't offer in the defense-focused Pro
• The Pro's integrated AI rating is about 98 TOPS, in line with last year's model, leaving an optional Nvidia RTX PRO 500 as its main AI upgrade

Getac has unveiled the B360X and B360X Pro, a pair of fully rugged 14-inch notebooks built around Intel's Core Ultra Series 3 "Panther Lake" processors and pitched at people who need to run AI workloads far from a reliable connection.

The Taiwanese company says the new machines deliver up to 180 TOPS of AI compute with both models due in November 2026 and no pricing announced.

Getac isn't subtle about who the Pro is for. Its product page calls it "purpose-built for defense and military professionals" and lists defense as its only industry, while the B360X is pitched at public safety, utilities, and industrial manufacturing. Our 2023 review noted that the B360 Pro is "sold as military-grade and mission-ready, out of the box," and that still seems true in 2026.

What has grown is the breadth of the push. In March, Getac launched CommandCore, a rugged drone ground control station that it pitches to "Army aviation and integrated air defense systems." Rivals are chasing the same customers, with Durabook's triple-screen Z14I-DX3 aimed at command posts and drone teams.

Plenty of upgrades under the hood even as focus remains centric to AI

Both notebooks replace the outgoing B360 and B360 Pro's 13.3-inch 1080p panel with a 14-inch, 1920 x 1200 LumiBond display rated at 1,400 nits, plus a touchscreen that Getac says works with compatible gloves or in the rain.

According to Getac's B360X spec sheet, the larger screen fits in the same 342 x 281mm footprint as the 13.3-inch models, with a slightly slimmer chassis and a 2.38kg weight.

Wi-Fi 7 and Bluetooth 6 are standard, with 4G LTE, 5G Sub-6, and a dedicated L1/L5 GPS module optional, alongside up to two Thunderbolt 4 ports, a smart card reader, TPM 2.0, a 5MP Windows Hello camera, and an optional second SSD for up to 4TB in total. The usual rugged credentials are included: MIL-STD-810H, MIL-STD-461G, and IP66 ratings; a 6ft (1.8m) drop rating while operating; and optional salt fog certification.

Intel calls the 180 TOPS figure platform TOPS, which adds the NPU, integrated graphics, and CPU together. Only Panther Lake's "X" chips reach it because only they carry the 12-core Arc B390 graphics, which accounts for 122 TOPS on its own, plus a 50 TOPS NPU, according to Intel's figures.

Getac offers exactly one such chip, the Core Ultra X7 358H, and its own footnotes confirm it is only available on the B360X. The B360X Pro tops out at the Core Ultra 7 356H and 366H, which use a four-core graphics version, and the 356H is rated at 98 TOPS on the same measure.

That is essentially level with last year's hardware. The Core Ultra 7 255H offered in Getac's 2025 B360 range is rated at 96 TOPS, so on Intel's own yardstick, the defense-focused model's integrated AI budget has barely moved.

A lot of AI muscle on offer

What has changed is where those TOPS live. The NPU jumps from 13 TOPS on the older Arrow Lake H chips to as much as 50 TOPS, clearing the 40 TOPS bar Microsoft set for Copilot+ PCs, and Getac now lists both models under that banner on its website. That matters to the target audience and the need to know it is achievable on battery, since NPUs are built around sustained, low-power inference.

Even that isn't new for the family. Getac billed its Lunar Lake-based B360 Plus as the first fully rugged Copilot+ PC when it launched in June 2025, with a 48 TOPS NPU, and the Core Ultra 7 258V it offers is rated at around 115 platform TOPS. When we covered that machine, we warned that the practical benefits "may not be as clear-cut as the branding suggests," and things have since come full circle.

Microsoft has also cooled on the NPU push. As PCWorld pointed out in January 2026, few applications make use of NPUs, and Windows' newer AI tooling lets developers run inference on GPUs and CPUs as well. Getac's generational AI claims hold up for the top B360X configuration, and for NPU-bound work across the range. For the Pro, they are a stretch.

Getac hasn't explained why it reserves the X7 for the B360X, but Intel's platform rules likely explain it. The X-series models seem locked to LPDDR5x memory, and Notebookcheck's spec comparison shows the X7 358H exposes 12 PCIe lanes, versus 20 on the Core Ultra 7 356H. A machine built around a discrete GPU option and an expansion bay needs those lanes more than it needs a bigger iGPU. This means Getac's stated specs might need a small tweak to note that the B360X also comes in an LPDDR5X-touting trim, since it makes no mention of any memory but "DDR5" so far.

The B360X Pro takes a different route, offering an optional RTX PRO 500 Blackwell with 6GB of GDDR7. Getac's release frames it as a graphics upgrade, but this is where the Pro's real AI headroom sits.

Nvidia rates the RTX PRO 500 at 294 AI TOPS, but its own RTX PRO line card carries a crucial footnote: "NVIDIA Blackwell GPUs are based on FP4 AI TOPS with sparsity." Intel quotes 8-bit integer (INT8) operations without that sparsity assumption, so the two numbers can't be compared directly.

For context, the RTX 5090's headline 3,352 AI TOPS corresponds to 838 INT8 TOPS without sparsity, a four-to-one ratio. Applying the same ratio to the RTX PRO 500 puts it at roughly 74 dense INT8 TOPS, below the 122 TOPS Intel claims for the X7's integrated graphics.

That doesn't make the B360X the better AI machine for every job. The Nvidia card brings 288GB/s of dedicated memory bandwidth, native FP4 support, and CUDA, still the default target for much AI software, although 6GB of VRAM limits the size of the model it can hold. Getac hasn't published benchmarks for either configuration.

Via TechPowerUp

Astell & Kern PD5: Two-minute review

It’s been a long time since Astell & Kern has had a genuinely affordable digital audio player in its line-up, and the launch of the PD5 does nothing to change that.

It’s been a while since the company had anything as visually unremarkable as this in its catalog, too — but try not to get hung up on the perceived value and instead concentrate on the priceless nature of its sound quality.

And besides, a considerable amount of money put Astell & Kern’s way for a PD5 may not buy you much in the way of design or tactility but it gets you an extremely thoroughly specified device. Eight Cirrus Logic DACs? 20-point EQ adjustment? Balanced and unbalanced headphone outputs? The ability to work as a desktop DAC? Enough power to get a tune out even the most uncooperative headphones? The PD5 has all of this, and more — and it’s all in the service of audio performance that’s as good as any portable device at anything like this price point.

When it comes to clarity, to detail retrieval and to outright fidelity, the PD5 is extremely hard to criticise and yes, it fully emphatically deserves a place in our best MP3 players guide. And its talents don’t stop there, not even close — its sound is punchily robust, effortlessly dynamic, brilliantly controlled and organised, and thoroughly engaging. If you take your portable listening experience seriously (or your desktop listening experience for that matter — the PD5 is also an excellent USB DAC), this Astell & Kern absolutely demands your attention.

Away from audio quality the PD5 isn’t perfect; there are rare occasions on which it can be glitchy in operation, its on-screen menus are approaching ‘miniscule’, and it neither looks nor feels quite as expensive as it actually is. All of this, though, is made to seem fairly trivial by the way the PD5 performs its primary function.

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Price and release date

• Priced $1,049 / £899 / AU$1,349
• Announced in August 2026; released in September 2026

The Astell & Kern PD5 digital audio player is on sale now, and in the United Kingdom it costs £899. In the United States it will set you back $1,049, while in Australia it goes for AU$1,349.

Anyone familiar with such devices knows that for an A&K player, that's actually not so expensive — again, note my 'for an A&K player' wording there. This company's flagship DAPs go for $3,999 / £3,799 / AU$6,599 (Astell & Kern A&ultima SP4000 and its newer T-suffixed sibling, I'm looking at you), but you can also get a hi-res player from A&K's sub-brand — called the Activo P1 — for $430 / £399 (around AU$869).

There are crucial differences in the specifications which I'll go into, but it's good to have an idea of the market at least, when looking at such a fee and being told by its makers that it is "entry level".

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Features

• Eight (count ‘em!) Cirrus Logic CS43198 DACs
• Balanced and unbalanced headphone outputs
• Bluetooth 5.3 with aptX HD support

The PD5 isn’t the first Astell & Kern product that’s been so thoroughly specified that discussing every one of its features in any kind of depth might take up most of your day. So let’s say right from the off that no eventuality has been overlooked where the PD5 is concerned, and then let’s deal with the headlines and then try not to get too bogged down in all the rest of it…

Never knowingly understated, Astell & Kern has fitted eight Cirrus Logic CS43198 DAC chipsets here. The user interface gives you the option of listening using four, or all eight — listening in octa-mode offers optimal sound quality, but eats into battery life quite badly (engage all eight DACs and you’re looking at 15 hours of playback as opposed to the 28 or so you should get when listening to ‘just’ four). Taking the battery from ‘flat’ to ‘full’ takes around two-and-a-half hours.

In terms of feeding digital audio content to that considerable digital-to-analog conversion line-up, you have a few options. The bare-bones Android user interface Astell & Kern has deployed here, along with dual-band wi-fi, allows you to download the app of your favourite music streaming service(s), and 64GB of internal memory allows you to store some content to the PD5’s local storage. There’s a microSD card slot on the bottom of the frame (next to the USB-C slot that’s used for data transfer and for battery charging) too — it can accept cards of up to 1.5TB, which should prove adequate for all the hi-res content you might desire.

The player can deal with resolutions of up to 32bit/384kHz PCM and DSD256, and is Roon Ready as well as featuring full-on Qobuz Connect integration. Or you might decide to bypass the fancy-pants D-to-A circuitry altogether and use Bluetooth to drive a pair of wireless headphones (although it does seem a bit of a weird idea to spend £899 on a player in order to use it like a smartphone).

(Image credit: Future / Simon Lucas)

The PD5 features Bluetooth 5.3 with codec support up to and including aptX HD. It also has two headphone outputs: an unbalanced 3.5mm socket and a balanced 4.4mm alternative that’s compatible with five-pole jacks. Use the 4.4mm output, select octa-DAC mode, and the Astell & Kern can summon a mammoth 8Vrms of power — so even the most demanding headphones will have to do as they are told. But there are quite a few ways in which the sound can be affected, both by the device and by the person using it, before it reaches these outputs.

The ‘EnviroTune’ feature uses a mic integrated into the top edge of the PD5 to analyse ambient noise in real time and correct those frequencies affected by it without just making everything louder. The ‘Teraton Alpha’ audio engine that’s familiar from some of Astell & Kern’s uber-pricey DAPs minimises noise from the power supply and creates amplification that’s as stable and distortion-free as possible. And ‘Enhanced Signal Alignment’ technology synchronises outgoing signals in an effort to negate phase distortion and deliver the most coherent sonic image possible.

The end user, meanwhile, gets a choice of digital filters and some quite in-depth EQ adjustment to look into. The default filter is ‘low latency fast’, and it is as distinct in its effect as ‘low latency slow’, ‘phase compensated fast’ and ‘phase compensated slow’ are from each other. A ten-band EQ offers fairly granular adjustment of the sound, and an ‘advanced’ EQ interface allows frequency response to be defined using a graph with 20 (!) plotting points.

The PD5 will function as a desktop too, thanks to its USB-C slot’s ability with data transfer. Simply make a hard connection to, say, your laptop, select ‘DAC input’ in the ‘USB Mode’ sub-menu, and the Astell & Kern automatically switches on octa-DAC mode in order to give your computer a proper go at sounding decent for once.

Features score: 5 / 5

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Sound quality

• Clarity, detail and dynamism in equal measure
• Organised and fully resolved sound
• Well-controlled punch and drive

The Astell & Kern PD5 is an admirably unpicky device, although obviously there are limits. Give it the sort of standard content available on a proper music streaming service (which means a CD-quality resolution of 16bit/44.1kHz) and it does an admirable job of maximising the information available to it. Step down from there, though, to the sort of squashed nonsense served up by Apple Music or Spotify, and the PD5 struggles to conceal its disdain. But you didn’t spend proper money to listen to compressed poverty-spec streams, did you?

So a 16bit/44.1kHz stream of Hood’s Western Housing Concerns is delivered in tremendous fashion. The presentation is open and spacious, with an expansive soundstage on which each element of the recording gets ample room in which to stretch out and express itself. The layout is rigorously observed, and the PD5’s powers or organisation mean the stage is locked in place and completely convincing.

Tonality is absolutely neutral, the player offering neither heat nor chill to the colour of the recording. There’s excellent clarity to the presentation, a really dark and quiet background in front of which the band perform — and at every stage of the frequency range, there’s absolutely fanatical attention to detail. No matter how minor an occurrence, how far back in the mix, or how fleeting, the Astell & Kern will seize upon it and give it precisely the amount of weight and emphasis it requires.

Frequency response is similarly even-handed, and similarly convincing as a result. A switch to a DSD64 file of Michael Jackson’s Don’t Stop ‘Til You Get Enough reveals deep, textured, tremendously varied and thrillingly solid low-frequency response, and bass is controlled to the point that rhythmic expression is never less than positive and confident. Midrange resolution is no less accomplished, with the singer’s mic technique made every bit as apparent as his tone and timbre. The top of the frequency range strikes a lovely balance between substance and brilliance, and again there’s an absolutely stack of information available. The PD5 doesn’t over- or underemphasise any part of the frequency range, and as a result this glossy, audibly expensive recording sounds as natural and polished as it ever did.

Dynamic headroom is considerable when it’s time to describe big variations in volume or intensity or attack, and the more subtle dynamics of harmonic variation and transient response are never overlooked. The entire presentation hangs together seamlessly, the Astell & Kern offering a proper sense of unity and singularity to recordings of every type.

And perhaps best of all, there’s nothing show-y or even especially analytical about the way the PD5 goes about things. As far as insight into a recording goes it’s basically untouchable at the money, but that’s not what defines the Astell & Kern. It’s the energy of its sound, the engaged and engaging nature of its performance, that makes it as entertaining a listen as you’ll get from a portable device costing less than four figures.

Sound quality score: 5 / 5

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Design

• 120 x 70 x 16mm (HxWxD)
• 4.1in touchscreen display
• 198g

There’s absolutely nothing wrong with the Astell & Kern PD5 as a physical object, let me make that clear from the off. The dimensions are nicely palm-sized, the weight is sufficient to make it feel like a premium product without being a burden to a jacket pocket, and the standard of fit and finish is impeccable.

And I’m not about to pretend I miss the painfully pointy edges of some previous Astell & Kern models, either — the PD5 is a much friendlier proposition in this respect — nor the painfully self-conscious offset screen of the old SR35. But there’s nothing remarkable about the design of the PD5, nothing to set it apart from the pack of products competing with it for your hard-earned. Even the volume dial, previously a little Astell & Kern design flourish, is unremarkable here.

I realise I can’t have it all ways, and a calm and sensible aesthetic is preferable to something overwrought and wilful. But I find myself slightly concerned that for the PD5, Astell & Kern may have bypassed its industrial design team altogether…

Design score: 4.5 / 5

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Usability and setup

• Bright, rather reflective screen
• In-depth menus
• Occasional operational oddities

As mentioned, the PD5 uses a stripped-back version of Android as its user interface — and it’s all the better for it. Too many digital audio players pointlessly include a lot of your smartphone’s functionality for no reason, as far as I can tell, other than to eat into the machine’s local memory.

The 4.1in touchscreen is bright, although rather more reflective than is ideal, and it’s responsive without being on a hair trigger. The Astell & Kern’s menus are extensive, and as a result the font size is really very small — but squint and you should be able to make out what’s what (even if the occasional use of ‘red on black’ for text is a terrible idea).

Setting up the PD5 to your satisfaction isn’t the work of a moment. But when you finally get finished you can at least be sure you have a digital audio player that’s primed to do exactly what you want, and in the manner you want it.

There are occasional glitches in operation, though — or, at least, that’s my experience. I install the Tidal music streaming app, and also insert a microSD card full of hi-res content. But every now and then, both sources will play at the same time. Pressing ‘play’ on a Tidal file usually stops playback from the local storage and vice versa — but not always, which is a weird one and no mistake.

Usability and setup score: 3.5 / 5

(Image credit: Future / Simon Lucas)Astell & Kern PD5 review: Value

Because the majority of the money you’ve spent on a PD5 has gone on the internals, it’s difficult to suggest its perceived value is especially high even if you accept that it’s beautifully constructed and even quite tactile. No, the value for money Astell & Kern is offering here doesn’t reveal itself until you start listening — and then it becomes quickly apparent that your money has been very well spent indeed.

Value score: 4.5 / 5

Should I buy the Astell & Kern PD5? Buy it if... 

You have needy headphones
The PD5 is equipped to make even the most difficult headphones bend to its will.

You’re after uncompromised sound quality
With the Astell & Kern you get what you pay for and then some.

You’re sharp-eyed as well as sharp-eared
Navigating those set-up menus is no job for anyone who’s misplaced their spectacles.

Don't buy it if... 

You want to see, as well as hear, where your money has gone
The PD5 is a perfectly pleasant little device, but its appearance doesn’t exactly shout “almost a grand!”

You don’t own decent wired headphones
The Astell & Kern sounds perfectly fine via Bluetooth, but really your smartphone will do a pretty similar job.

Astell & Kern PD5 review: Also consider

FiiO M23
The FiiO M23 has been around for a couple of years now, but really this only means it’s now competing almost directly with the PD5 on price. It remains a biggish, bulky-ish device, it’s true, but it also remains one of the most poised, direct and downright musical digital audio players around. Give it the right stuff to work with in terms of content and headphones, and the sound quality is enthralling. Read our full FiiO M23 review

How I tested the Astell & Kern PD5

• Listened using Tidal and Qobuz files
• Popped a 512GB microSD card loaded with extremely hi-res content into it
• Used Sennheiser IE900 and Meze Audio Strada for most of my testing

I installed the Tidal music streaming app and also deployed Qobuz Connect (it’s a slicker experience altogether than installing the app onto the player), and I inserted a 512GB microSD card loaded with extremely high-resolution content.

Then I connected a pair of Sennheiser IE900 IEMs using the 4.4mm balanced output, and also used a pair of Meze Audio Strada headphones connected the same way.

I also investigated the PD5’s desktop DAC credentials by plugging it into an Apple MacBook Pro and using the Qobuz and Tidal desktop apps to listen to hi-res using the same pairs of headphones.

• First reviewed in October 2026

• Read TechRadar's reviews guarantee

Now that I've seen and touched the touchscreen MacBook Pro, I'm starting to understand that, despite the invention of a Frankenstein-ed new display term ('PROLED'?), this display might be something special. It looks extraordinary, and when compared visually with, say, the Lenovo Yoga Pro 9i Aura Edition, which also has a Tandem OLED, it doesn't even seem to be of the same species.

There are good reasons for this, which span from the fabrication process to the core specs of what is clearly Apple's most advanced laptop display yet.

Starting with manufacturing, Apple says it employed an unnamed manufacturing partner to build these panels in what's known as a Gen-8 OLED Fab. Seeing as there are just four of these cutting-edge facilities in the world, and that one of them is operated by Samsung Display, which has reportedly partnered with Apple on display tech previously, I'd guess it's Samsung.

One of the most notable features of these manufacturing facilities is their massive size. Whoever the partner is, we know that the touchscreen MacBook Pro's panels are being manufactured in a chamber around the size of the Colosseum in Rome. These huge facilities improve yield, efficiency, and costs.

Though less than a millimeter thick, the laptop PROLED XDR panel is comprised of 50 layers. Even what appear to be individual panels have many layers, sometimes just a nanometer thick. The two stacks of OLED have, for instance, 30 layers, though the organic self-emissive material is layered onto the panels and dried, making the layers range from one nanometer to hundreds of nanometers thick. FWIW, I held some of these individual panels, and they are incredibly thin and light.

Behind the OLED is the TFT (thin-film transistor) glass substrate that powers and drives the imagery on the surface.

What's not mentioned here is the digitizing layer (basically the same touch technology you find on the iPhone and iPad) that drives touch; that's because it's integrated with the whole in a process that, if it isn't already, probably should be patented.

(Image credit: Lance Ulanoff / Future)

Despite working with a third party, the production is another Apple trademark bespoke process, and one not necessarily replicated in other Tandem OLED or even basic OLED display manufacturing processes.

The specs bear that out. Where the Lenovo Yoga Pro 9i Aura Edition touch display has a pleasing 1,000,000:1 contrast ratio, Apple's touchscreen MacBook Pro boasts 10,000,000:1. In person, and even under bright lighting conditions, the blacks and definition I saw on the touchscreen MacBook Pro PROLED XDR are stunning.

The MacBook Pro display also looks better because Apple equipped it with up to 2,000 nits for HDR and SDR content, and both an anti-reflective screen and the option of the nano-etched one, as well. Essentially, I could see myself reflected in the Lenovo, and could not detect my reflection in the MacBook Pro.

Also, for those concerned about fingerprints collecting on the touchscreen MacBook Pro, Apple has applied an oleophobic coating to the standard and etched displays. In my hands-on time with several of the new laptops, I didn't notice any significant fingerprints.

Lance Ulanoff / FutureLance Ulanoff / FutureLance Ulanoff / Future

Compared to the Lenovo Yoga Pro 9i's 1,600 nits, the touchscreen MacBook Pro looks better equipped for working in direct sunlight. Based on what I saw, the MacBook Pro also maintains much better off-axis color quality.

Looking beyond direct visual comparisons with pro-grade Windows laptops, Apple is positioning the touchscreen MacBook Pro display as both a color reference screen and one that has grayscale precision reportedly sufficient for medical diagnostics. This is basically a reference monitor on a laptop frame.

So, sure, the story is that we finally have a touchscreen MacBook, but it may be the visual quality that ultimately wins over creative, business, medical, and film professionals. It's probably the best display Apple has ever produced, not just for a laptop, but possibly for any Mac system.

Of course, a screen panel alone does not make a laptop, and there are questions around whether everyone wants or needs touch, if the redesigned hinge is stable enough to keep the laptop from wobbling with every touch — and if touch capability is worth the almost $1,000 premium.

“Captain Picard stood for the best of us,” wrote William Shatner, who played Captain James T. Kirk.

• Saab develops a working underwater drone demonstrator in less than a year
• Sweden's new underwater vehicle has completed its initial sea trials
• Military authorities will evaluate the drone before deciding on procurement use

Saab has built a large uncrewed underwater vehicle demonstrator, known as LUUV, in collaboration with Swedish defense authorities.

The company finished development in less than a year after receiving its contract, and the prototype has already passed its first sea trials successfully.

This means the drone now counts as a working demonstrator, giving engineers a real platform they can test and improve in the water.

Quick progress and real-world testing

The Swedish Defence Materiel Administration (FMV) and the Swedish Armed Forces are also involved, and they will help judge whether the technology suits military needs.

Saab says the fast progress came from officials and engineers working side by side to develop and test the equipment in a short period.

“This project demonstrates how new technologies can be rapidly translated into practical operational value,” said Mats Wicksell, head of Saab’s business area Naval.

“The fact that we have also been able to develop and test the system in such a short timeframe demonstrates the strength of the collaboration between the Swedish Armed Forces, FMV and Saab.”

Saab argues this kind of partnership can shorten the time between spotting a military requirement and testing a practical solution at sea.

The completed sea trials offer a first look at how the vehicle performs in actual ocean conditions, rather than only during development testing.

A demonstrator is an early working model built to prove that an idea functions, rather than a finished product ready for regular service.

However, this milestone applies only to a demonstrator, so further development is still needed before the system can be considered ready for operational use.

Going forward, Saab expects the next phase to bring more autonomous functions, software improvements and additional sensors.

Software, benefits, and open questions

The project relies on Saab's Autonomous Ocean Core, a proprietary software platform designed to let underwater and surface vessels operate autonomously without constant human control.

This software-focused approach lets developers build automated functions into marine platforms and adjust them later as operational requirements change over time.

Saab believes uncrewed underwater vehicles could extend mission endurance, improve awareness of activity beneath the surface and strengthen a navy's overall capabilities under the sea.

Such platforms could prove useful for maritime surveillance and other underwater missions, although Saab has not shared the full specifications of this particular demonstrator.

The company has not announced the vehicle's size, maximum depth, endurance, speed or payload capacity, so many important details remain unknown for now.

The next development phase will therefore be important, because it should show which capabilities the platform can actually deliver to the armed forces.

Swedish authorities will, however, use these results to decide whether the technology deserves more investment.

Whatever happens next, Saab has shown that close collaboration between industry and government can deliver a working underwater vehicle in a remarkably short time.

Via Drone Front

I did not expect to love the AirPods Max 2 so much, upon testing them. Nobody is more surprised than me that I chose Apple's new over-ears for Headphones of the Year at the TechRadar Product of the Year Awards 2026.

Bowers & Wilkins has won this award the last few years, and I even hovered over the excellent five-star HEDD HEDDPhones D1 for a moment. But despite being trailblazing cans from a pioneering new audio outfit (and for some, that will fit the bill perfectly) the D1 are open-back, wired, passive beasts. And the thing is, AirPods Max 2 just do so much more for the money.

Even though you, the readers, also voted the AirPods Max 2 for this award in our TechRadar Readers' Choice Awards 2026, I'm certain there will still be some cries of discontentment about this. 'Wait, so Apple's AirPods Max 2, the set that look almost identical to the 2020 originals, and have lower battery life than any of the competition, and have no wireless hi-res codecs (and the wired lossless support also came to first-gen model anyway), and that still come with the weird headphone bra case, those won?'

Yes. Yes really. And if you give me a few minutes, I'm going to explain why. That said, if you still want to comment below, I'm all ears.

(Image credit: Future)Yes, AirPods Max 2 look similar to their 7-year-old siblings — they're just that good

OK, so the lossless stuff with a wired connection isn't new, and neither is the look of the things. But what has changed with these March 2026 issue cans is the inclusion of a new H2 chip and all of its computational wizardry. And this makes them punchier sonically, ups the ANC efficacy significantly (and it wasn't at all bad beforehand, remember), and adds new perks such as Live Translation, Adaptive Audio, Conversation Awareness, Head Gesture control and the Camera Remote feature.

The one caveat is that some features require an iPhone 15 Pro or newer. Android owner? This isn't your day I'm afraid, and I'm sorry about that — your closest pick would be last September's Bose QuietComfort Ultra Headphones (Gen 2).

But the box-fresh Apple Intelligence features genuinely make for an excellent, full-throttle music and call-handling experience that does everything you need and does it with minimal fuss, app consultations, screen taps or button pushing.

For me, the right way to go about Apple's incredible implementation of ANC here is to deploy both noise cancellation and the Conversation Awareness toggle, which you'll find with a tap of the AirPods Max bubble near the top of the settings tab on your iPhone, or by swiping down from the top-right of your screen, to get the Control Center.

What it means is that by default, you will not be interrupted, but when you speak up, music is automatically lowered and external noises are piped in as you need them. And I'm mentioning this specifically because so many rival headphones aim to offer this (ie. a cocoon of silence as a backdrop for your music, but with a decent bit of software that kicks in when you speak to take it all away and pipe ambient sound in) and don't get anywhere near achieving what Apple has done.

OK so it's not groundbreaking. But it is so, so good in your everyday work life — I implore you to try it, if you get the chance.

That said, if you take just one thing away from this particular feature let it be that the claim in the marketing release is true: the noise-cancellation in AirPods Max 2 really is 1.5 times better than before. It's outstanding.

These headphones are almost deviously, head-shakingly good at isolating and melting away the noise from my surroundings, but without the wind-tunnel effect or pressure sensation I have occasionally found disconcerting or nauseating in other cans — the handful that offer ANC anywhere near this effective, at any rate.

I maintain that for emotive music or movies using Apple's special (and decidedly specific) brand of head-tracked Spatial Audio with a backdrop of near-silence wherever I tested them, Apple's AirPods Max 2 went straight to the top of a big pile.

Did I mention how comfortable they are? So many headphones start to dig in across the top of my head. That never happens with Apple's slightly odd-looking but delightfully nonrestrictive mesh headband.

Simply put, they're the headphones I started reaching for on the daily since I acquired them in March of this year, and they're still the set I reach for now. I didn't think this would happen, but it would be wrong of me to deny it, all the same.

Despite several other new releases resting on my desk at any given moment, it's the AirPods Max 2 I always go for when the working day is done. And I was taught a long time ago that this can only mean they're the best product you've currently got — we cannot argue with our own subliminal preference. I mean, that is as long as they haven't run out of battery…

CEO Jensen Huang may have been the target of Chinese espionage activity, The New York Times reports.

Archive of Our Own, one of the world’s biggest fanfiction sites, went down hard.

• Report says AI is being given too much authority and autonomy to handle business-critical security decisions
• 70% of organizations delegate decisions to AI that would have been made by a human 12 months ago
• A “control gap” exists between the authority given to AI and an organization’s ability to audit it

Delegating complicated decisions to AI is leading organizations to place relatively new technology in charge of user permissions, in control of sensitive data, and risk security issues, new research has claimed.

A study by Illumio looked into how organizations are handing decision recently performed by humans to AI, without apparently taking into account the potential for serious issues to arise.

It found the majority (70%) of organizations are delegating previously human-made decisions to AI, but only around half (49%) can see what action is taken by an AI agent an audit it.

AI recommendations

The nature of AI agents makes this a particularly risky strategy, and the research has found that an incredible 56% of organizations are willing to accept greater cyber risks than 12 months ago. Almost half of that group (47%) have admitted to permitting AI actions in production environments.

Allowing AI agents to have free run of enterprise data, applications, and infrastructure is leading to wholly avoidable situations. Illumio’s research found that a control gap between the rules that organizations believe AI agents are operating within and what the technology is actually capable of.

Some 76% of organizations have experienced one or more security issues thanks to AI in the past year, with 35% revealing that following AI recommendations would result in a security risk. Meanwhile, 26% reported AI agents or tools had connected to other systems without the necessary approval.

More worrying is the 24% of organizations reporting AI’s drive to access data it shouldn’t be able to see, and another 24% stating it has triggered (or produced) an unauthorized action.

Strict limits

Of course, AI is already used within cybersecurity, and has proved successful with multiple vendors rolling out AI-enhanced solutions. But while 67% of organizations are using AI cybersecurity tools, 32% let AI conduct investigations without human oversight and 29% permit the software to deploy code changes completely unsupervised.

“The biggest AI risk may not be a system that breaks its rules. It may be one that was never given meaningful limits in the first place," noted Raghu Nandakumara, VP of Industry Strategy at Illumio.

Unleashing AI agents without constraints has left businesses struggling to control the software.

As Nandakumara observes, AI agents are persistent, and attempting to outwit their next move is pointless. Instead, “businesses must set strict limits on what they can access and do, then enforce those limits within the application and across the underlying infrastructure. That is how you contain unexpected behavior before it causes real damage.”