TechRadar News

• Tbps-scale DDoS attacks have shifted from rare anomalies to constant threats
• Hacktivist groups weaponize automation and botnets to destabilize fragile infrastructure
• Political disputes increasingly spill online, triggering destructive waves of cyber aggression

The first half of 2025 marked another MAJOR escalation in distributed denial-of-service (DDoS) activity, with new NetScout research documenting more than eight million attacks worldwide in these six months.

More than three million attacks were recorded across Europe, the Middle East, and Africa, underscoring the regional strain.

It also noted terabit-per-second scale strikes, once rare anomalies, have become almost routine, with peaks reaching 3.12Tbps in the Netherlands and 1.5Gbps in the United States.

Political conflict drives digital aggression

These findings suggest DDoS attacks are no longer an occasional disruption, but an entrenched method of destabilizing essential networks, as geopolitical tensions remain a key trigger for major attack campaigns.

NetScout noted how disputes between India and Pakistan spurred extensive waves of hostile activity against Indian financial and governmental systems.

Similarly, during confrontations involving Iran and Israel, over 15,000 strikes targeted Iranian infrastructure in a matter of days, while fewer than 300 targeted Israel.

Even international forums were not spared, with events in Switzerland experiencing more than 1,400 incidents in a single week.

Much of this scale also relies on compromised devices operating as botnets.

In March 2025 alone, attackers launched an average of 880 botnet-driven incidents daily, with peaks of 1,600.

The compromised systems typically included routers, servers, and IoT devices, often relying on known flaws rather than undiscovered vulnerabilities.

Despite years of security warnings, these weaknesses remain consistently exploited, enabling short but impactful campaigns that disrupt dependent services.

For organizations relying only on basic antivirus or endpoint protection, such sustained botnet traffic presents challenges that overwhelm conventional safeguards.

Furthermore, the evolution of DDoS campaigns has been accelerated by automation and artificial intelligence.

Multi-vector strikes and carpet-bombing techniques now occur faster than defenders can respond, creating asymmetric pressure.

NetScout also pointed to the emergence of “rogue LLMs,” which provide hostile actors with accessible planning and evasion methods.

Combined with DDoS-for-hire platforms, these tools have significantly reduced the barriers for inexperienced attackers, enabling high-capacity strikes with minimal technical depth.

The outcome is that Tbps-scale incidents have shifted from rare spectacles to constant risks.

Among hacktivist collectives, NoName057(16) continues to execute the most frequent campaigns, far outpacing rivals.

In March, the group claimed more than 475 attacks, primarily directed at government portals in Spain, Taiwan, and Ukraine.

Their reliance on varied flooding techniques indicates both coordination and persistence, suggesting ideological motivations beyond opportunistic disruption.

While new players such as DieNet and Keymous+ entered the scene with dozens of attacks across multiple sectors, their activity still fell short compared with NoName057(16)’s scale.

“As hacktivist groups leverage more automation, shared infrastructure, and evolving tactics, organizations must recognize that traditional defenses are no longer sufficient,” stated Richard Hummel, director, threat intelligence, NetScout.

“The integration of AI assistants and the use of large language models (LLMs), such as WormGPT and FraudGPT, escalates that concern. And, while the recent takedown of NoName057(16) was successful in temporarily reducing the group’s DDoS botnet activities, preventing a future return to the top DDoS hacktivist threat is not guaranteed.”

You might also like

• The best antivirus software can keep your devices protected
• Surf the web at speed with the best browser
• AI in the workplace: the legal risks businesses can’t afford to ignore

From smart watches to smart supply chains, digital transformation is reshaping how we live, work and interact. But the true vision of an interconnected world – where billions of everyday items are embedded with intelligence – remains just out of reach.

That’s not because the ambition is too bold. In fact, it’s widely recognized. McKinsey estimates the Internet of Things (IoT) could deliver up to $12.5 trillion in global value by 2030. Extend that connectivity to low-cost everyday objects, and the potential increases dramatically.

So why hasn’t it happened yet?

The biggest roadblock is supply. A truly connected world needs a colossal volume of chips. But we’re not talking about the advanced processors that power smartphones or autonomous vehicles – we’re talking about simpler, highly distributed, low-cost chips that can be embedded into everyday items.

These so-called ‘legacy’ chips – typically manufactured on 28nm or larger nodes – are anything but outdated. They’re critical to modern electronics, supporting high-performance processors and underpinning countless devices across consumer, industrial and automotive markets.

We saw just how vital they are during the COVID-19 pandemic. Shortages of these chips brought global manufacturing to a standstill, delaying everything from cars to home electronics. The ripple effects were felt across entire economies. As the number of smart devices scales, how to prevent a reoccurrence is an ever-present consideration.

So why not just build more chip factories?

Traditional fabs can’t meet future needs

Scaling legacy chip production is incredibly difficult. The equipment is aging and hard to replace. New fabs cost tens of billions and take years to build. And since mature-node chips don’t deliver the same profit margins as cutting-edge silicon, there’s little financial incentive to invest.

Even worse, legacy fabs are environmentally demanding. They consume vast amounts of energy and water. While next-gen fabs are built for sustainability, retrofitting legacy fabs to reduce their carbon footprint is expensive and complex.

To unlock true scale, we need a fundamentally different approach: one that’s faster, affordable, more flexible – and designed for sustainable, high-volume production.

Enter FlexICs

Flexible integrated circuits (FlexICs) are a new class of semiconductor: a chip that’s ultra-thin, physically flexible, and radically more sustainable.

Instead of traditional silicon wafers and high-temperature processing, FlexICs use thin-film technology and a low-energy manufacturing process. Think: less water, less energy, fewer harmful chemicals – and much faster turnaround.

This breakthrough unlocks a faster, more agile development cycle. Designs go from tape-out to volume production in weeks, not months. Costs are dramatically reduced. And it becomes viable to prototype, iterate and customize chips without the high stakes typically associated with silicon design.

Built for ubiquitous intelligence

This isn’t just a manufacturing story. It’s a transformation in how and where intelligence can be deployed.

FlexICs enable connectivity in places where traditional chips simply can’t go. Their ultra-thin, flexible form factor means they can be embedded directly into products or packaging, even on curved surfaces, delivering smarter, more connected experiences almost anywhere.

And they support standard communication protocols such as NFC, unlocking seamless, item-level intelligence for mass-market products – and the fastest digital connection between brands and their customers.

This embedded intelligence has multiple roles to play, from driving loyalty programs and exclusive content based on location or season, to product authentication, provenance information – or even facilitating improved reuse and recycling for a more circular economy.

It also offers a practical route for embedding real-time data collection into previously passive environments, whether that’s logistics chains, recycling systems, or agricultural operations.

Wherever you need intelligence, FlexICs bring it – sustainably, affordably, and at scale.

The future is flexible

But its goal isn’t about competing with silicon fabs. It’s about complementing them – addressing the unmet need for scalable, cost-effective, environmentally conscious production of the chips that will power the next wave of smart, connected systems.

We’re standing at the threshold of a hyperconnected future. But realizing that future depends on embedding intelligence at unprecedented scale. Flexible chips are the key to unlock that potential and make the vision of an interconnected world a reality.

We list the best performance management software.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

Marvel Rivals patch notes arrive every couple of weeks or so, bringing bug fixes, Hero balancing changes, and more general updates. Sometimes, as is the case with the most recent patch, skins are added, and certain Heroes receive buffs or nerfs to better bring them in line with the ongoing meta.

Season 3.5s latest patch notes are now live, and they focus on adding in new skins. There's some new details on Season 3.5's Hero Balancing as well. These constant changes to Marvel Rivals help keep Competitive Mode fresh while providing a steady stream of reasons for players to check back in regularly. Since launch, Marvel Rivals has really proven itself as a worthy contender, landing on our list of the best free games and the best crossplay games.

Here are the latest Marvel Rivals patch notes. These add new skins and an event as part of Season 3.5.

Marvel Rivals next patch notes

(Image credit: NetEase)

The next Marvel Rivals patch arrives August 28, and adds a new skin for Scarlet Witch, as well as a new event. Here's what's been confirmed so far:

• New skin: Phoenix Chaos (Scarlet Witch)
• New event: Earn 2x Chrono Tokens from Weekly Missions, Challenges, and Season Missions with up to 2000 bonus Chrono Tokens. Available until September 12.

Marvel Rivals latest patch notes

(Image credit: NetEase)

The latest patch for Marvel Rivals launched August 21 UTC. It added the Klyntar: Throne of Knull map and mode.

Marvel Rivals Season 3.5 Hero balancing changes

(Image credit: NetEase Games)

Below, you'll find a list of the latest balancing changes (August 21) in Marvel Rivals, detailing the buffs and nerfs that've been applied to each character.

• Blade (Buff) - Increase Lifesteal gained during Bloodline Awakening from 60% to 65%. Reduce self-healing reduction from 40% to 35%
• Human Torch (Nerf) - Reduce Flame Field (right-click on terrain) Damage Over Time from 25/s to 20/s; Reduce Pyro-Prison Damage Over Time from 50/s to 40/s
• Loki (Nerf) - Reduce Mystical Missile damage dealt by Loki and his illusions to enemies from 30 to 25. (Healing for teammates remains unchanged.)

As you can see, the latest Marvel Rivals Hero balances apply to Duelists and Strategists, upping their damage, or making their abilities more useful. You can read the full details here.

As a Vanguard player, I'm very interested in the recent Doctor Strange buff. I played him as a main during Season 0, and Season 1. Since then, I've mostly been using Magneto, but the nerfs could mean he's not as viable this season.

You Might Also Like...

• Here's when to expect the next season of Marvel Rivals
• Marvel Rivals Crosshairs explained
• Marvel Rivals Mutant list: here's all of the playable Mutant Heroes

• Parallels Desktop 26 launches with support for macOS 26 Tahoe and Windows 11 25H2
• The software introduces Internet Speed Booster, new IT management tools, and updated compliance
• Parallels continues its subscription model without a perpetual license option for customers

Cross-platform working should now be a lot easier folllowing the release of Parallels Desktop 26, the latest version of the popular virtualization software for Mac.

The update introduces compatibility with macOS 26 Tahoe and Windows 11 25H2, as well as a range of tools aimed at individual users and IT teams.

Among its new features is Internet Speed Booster, a mysterious addition with few details available on what it actually does, but its presence may catch the attention of users who run bandwidth-heavy tasks inside virtual machines.

macOS 26 changes

“With Parallels Desktop 26, we’re delivering compatibility with Apple’s latest macOS, and powerful new tools for the IT teams and developers who rely on Parallels every day,” said Prashant Ketkar, Chief Technology and Product Officer at Parallels.

“This release reflects our ongoing commitment to Apple innovation and our focus on enabling enterprises and IT admins to securely manage and scale their Mac environments.”

Parallels Desktop 26 improves how Windows virtual machines read Mac disk space.

Microsoft’s OS can now display the storage actually available on the host Mac, reducing potential errors during large software installations or updates.

The software also takes into account how macOS 26 handles background processes, allowing features like Coherence Mode to continue to run without interruption.

The update also adds new centralized management options for enterprise users. Administrators will soon be able to apply policies that control settings such as shared folders, USB access, and clipboard sharing.

The software has renewed its SOC 2 Type II compliance, which supports companies that have to meet strict internal and external security standards.

There’s a new guide for deploying single Windows applications on Macs without requiring users to run a full Windows environment, and integration with Jamf Pro for monitoring and managing Windows updates within virtual machines.

Scripts provided by Parallels will allow IT teams to run scans or start installations remotely.

Parallels Desktop 26 is available in Standard, Pro, Business, and Enterprise editions, but it continues to be offered only as a subscription.

The lack of a perpetual license option will no doubt be a disappointment for many, especially long-term users.

You might also like

• We've rounded up the best free WinZip alternatives around
• Democratized cybercrime: a new lower bar for hackers and higher stakes for security
• Millions of Brother printers threatened by multiple serious vulnerabilities

PDFs are everywhere in enterprise workflows – contracts, invoices, policy documents, audit trails. And yet, they’re rarely considered a security risk. That’s a problem.

Across industries, there remains a persistent belief that PDFs are static and inherently safe. This assumption creates a blind spot – one that’s increasingly at odds with how these documents are actually used.

PDFs today are often dynamic, shared across borders, embedded into automated processes, and packed with sensitive information.

Treating them as nothing more than electronic paper is not only outdated – it’s dangerous.

The Myth of the Safe PDF

In many organizations, document workflows haven’t kept pace with broader digital transformation efforts. While teams embrace AI, automation, and cloud services elsewhere, they often rely on legacy document management software that lacks visibility, auditability, and modern security features.

This complacency creates real risk. Sensitive data is moved, stored, and signed without appropriate controls. Metadata is overlooked. Documents are emailed instead of securely shared. And because PDFs don’t “feel” like a security threat, these behaviors often go unchallenged.

What’s more, compliance demands are tightening. Regulations such as GDPR, ISO 27001, and eIDAS 2.0, have raised the stakes, making poor document governance a potential legal liability. Inconsistent handling of digital files – even when unintentional – can result in significant exposure.

The Threat Landscape is Changing – Fast

In 2025, cyber threats are evolving faster than ever. Organizations now face average of over 560,000 new malware variants daily, with PDFs increasing being exploited as delivery mechanisms.

Meanwhile, AI-powered attacks have drastically reduced the time it takes for bad actors to move through an organization's systems. Many now breach, escalate, and extract data in under an hour. This accelerated threat landscape makes “invisible” gaps like unsecured PDFs even more dangerous.

Security Must Be Built into the Workflow

The answer isn’t just to bolt on security – it’s to embed it. Document protection should start at the point of creation and extend across the entire lifecycle. That means using tools and processes that allow for encryption, permissions control, version tracking, secure collaboration, and verifiable digital signatures as standard.

It also means making compliance easy, not optional. Security shouldn’t be something users have to remember – it should be part of how they work. The goal is to integrate protection in a way that supports productivity, rather than hindering it.

There’s a growing role here for AI and automation. From summarizing content and redacting sensitive fields to powering intelligent search and flagging anomalies, modern technologies can help organizations manage documents faster – and more safely.

The Opportunity for Partners and Providers

This shift has major implications for the channel. We’re moving into an era where partners are no longer just selling software licenses – they’re becoming workflow consultants and risk advisers. The organizations that stand out will be those that help clients reassess document handling as a core part of their cybersecurity and compliance strategies.

That means understanding sector-specific regulations. It means recommending secure-by-design tools that don’t rely on user vigilance alone. And it means being able to demonstrate how good document practices reduce liability, speed up processes, and build long-term resilience.

For partners who get this right, the rewards are clear - deeper client relationships, stronger value propositions, and a chance to lead in an area that many are still overlooking.

It’s Time to Bring PDFs into the Security Conversation

Too many enterprises still treat documents as an afterthought – something to deal with after the “real” cybersecurity work is done. But in reality, documents are where business happens. They’re where data is agreed, signed, stored, and exchanged.

And in today’s environment, where compliance burdens are increasing, AI is supercharging attacks, and breaches can escalate in minutes, PDFs simply can’t be left out of the equation.

If your organization is modernizing its infrastructure, reviewing its compliance position, or assessing security posture, then document workflows need to be part of that conversation.

Because if your PDFs aren’t secure, your business isn’t either.

We list the best cloud document storage.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

In the age of artificial intelligence, speed and automation are the key pillars of any system, yet they are no longer enough. Today’s customer demands both digital efficiency and emotional intelligence. They expect support that aligns with their needs in real time but also reflects a sense of humanity and care.

Customer experience (CX) leaders need to realign their vision with the changing dynamics of customer behavior. The challenge is no longer whether to implement automation, but how to design intelligent experiences that drive meaningful engagement. Seamless interactions are now expected everywhere, and personalization is not a differentiator - it is a must.

The customer has changed. Has your CX architecture evolved with them?

Consumer expectations have rapidly outpaced the capabilities of many legacy systems. Our recent report has found that 60 per cent of customers now prioritize minimal wait times, and 59 per cent say their preferred channels shift depending on context.

Customers interact across multiple touchpoints and are increasingly guided by emotion. They want convenience for routine tasks, but in moments of stress or urgency, they turn to human support. So much so, that 50 per cent of customers will abandon a brand entirely after just one negative interaction. That makes CX not just a service function, it is a business-critical risk factor.

Human Connection Remains Irreplaceable

Empathy cannot be automated. It is what transforms a customer support interaction from transactional to experiential. Today, customers specifically prioritize human connection over response speed, particularly in complex or emotionally charged scenarios. No artificial intelligence (AI) model, no matter how advanced, can replicate the emotional nuance of a live agent in those critical moments.

Voice support remains dominant for a reason. It is not only familiar but also effective, especially when digital channels fall short. It continues to be the preferred channel across demographics, particularly for Baby Boomers and Gen X. That preference spikes when the issue is sensitive, urgent or high value.

AI can enhance, but not replace, the human layer. The report found that 72 per cent of consumers are open to AI-driven interactions, but only when escalation to a human is easily available. This signals a need for thoughtful orchestration, not a wholesale automation agenda.

Trust remains the core barrier to adoption. While AI capabilities are evolving rapidly, public trust is still catching up. As with digital banking, full adoption will take time, and likely a generational shift.

Let users self-select into chatbot interactions when convenient. But make the path to human assistance frictionless and visible. Only a symbiotic relationship between machine intelligence and human empathy can produce the kind of experience that sustains long-term loyalty.

Personalization Is the New Loyalty Driver

Customers expect to be known, understood and remembered. In an age of data ubiquity, they view personalization not as a value-add, but as an obligation. With so much behavioral and transactional data at their disposal, brands have the tools to deliver tailored, predictive support. Yet, it is important to use it wisely.

Millennials, in particular, are willing to share personal data in exchange for better outcomes. That opens the door for proactive service and adaptive support strategies that evolve with the customer lifecycle.

Intelligent CRM systems and AI-infused agent assist tools can surface relevant context and enable personalized interactions at scale. Conversation histories can be retained across channels. Agents can be empowered in real time with insights into intent, sentiment and journey stage. The result is a frictionless handoff, even in an omnichannel environment.

The New CX Must Be Hybrid

The future of customer experience lies in hybrid orchestration. That means deploying advanced technology to handle high-frequency tasks, while preserving human bandwidth for high-emotion or high-value interactions. It is not about replacing people, it is about making them more effective. It is no AI that will replace people, it is people who use AI.

Here are five strategic imperatives for CX leaders navigating this transformation:

• Deploy AI to augment agent performance with real-time context, behavioral cues and next-best-action guidance.
• Ensure seamless fallback to human support is available at all digital entry points.
• Invest in empathy training for support staff, supported by full access to customer history and intent signals.
• Prioritize intuitive self-service design, but always offer a human escape hatch.
• Monitor journey satisfaction and emotional cues, and not just resolution time or deflection rate.

Winning in CX today is not about choosing between human and machine. It is about designing for both and orchestrating the handoff with precision. Empathy and intelligence must co-exist across the customer journey.

This is not just about keeping up with digital trends. It is about building a support model that earns trust, delivers value and strengthens customer relationships with every interaction.

This is how leaders stay ahead.

We list the best live chat software.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

We've all had those health moments when we're just feeling… meh. Nothing's exactly wrong, but you know you're not feeling your best and that you can do better – you just can't find the gumption to do so.

That was me after a recent bout of the flu completely ruined my routine. I wasn't eating great, I'd stopped reading as much as I'd like to, and I found myself choosing to stay up late rather than doing the smart thing and heading to bed. This, of course, created a snowball effect where I'd then be too tired to get up early to walk the dog or make a proper lunch.

I needed a little help and, in my experience, there are few better motivators than spending money. That was the reason I bought my first fitness tracker (the Fitbit Flex 2, back in 2017), which put me on a journey of smartwatches and trackers that eventually landed me with the Apple Watch SE in 2023.

But, after not too long, my SE was starting to feel stale. I’m not too taken by the smartwatch features on offer, I don't love the 'ring' fitness system and I've found myself increasingly annoyed with daily charging. And, personally, I can't wear a regular timepiece on one wrist and my Apple Watch on the other.

I was intrigued, then, when I discovered the Whoop band, which looked like it would address a lot of those gripes quite specifically. It's a screen-free, low-profile device that focuses on health and fitness, and has a multi-day battery. It also comes with a month-long free trial of its Peak membership which includes the Whoop 5.0 device itself (with a 14-day battery) and advanced tracking like Healthspan – which calculates the age your body is performing at (compared to your actual age) and gives you the data you need to improve.

I was slightly intimidated, because this class of fitness device is also usually attached to multi-million-dollar athletes and Olympians. But if you don't love the device during this trial period, all you have to do is pay the shipping cost to return it; which I had little doubt I'd be doing.

Whoop's states its mission is "to unlock human performance and healthspan", so it's for athletes like Virgil Van Dyke, James Tedesco, Michael Phelps, Rory McIlroy and Cristiano Ronaldo trying to maximize their potential – not me.

But then I strapped one on my wrist.

The real cost of a personal trainer

The Whoop 5.0 has overtaken the Apple Watch SE as my daily wearable (Image credit: Max Delaney / TechRadar)

Other than thinking it was too health and fitness focused for me, one of the biggest reasons I thought Whoop wasn't for me is its price. Whoop is a subscription service and without an active membership the band itself is useless.

You can subscribe to any of three different subscription options – One, Peak and Life. If you’re on the fence about whether Whoop is for you (don't worry, I certainly was), you can get a 1-month free trial of the Peak plan when buying direct from Whoop – which is what I did.

The Peak membership includes all the benefits of the more basic One tier. This includes the Whoop 5.0 band – which now boasts an improved 14+ day battery over the Whoop 4.0 – sleep/strain/recovery insights, personalized coaching, heart-rate tracking and women's hormonal insights, plus advanced features like Healthspan and a Real-Time Stress Monitor that analyzes how much you've been stressed based on your heart rate outside of physical activity and offers reasons why.

You also have the option to pick up Whoop on Amazon – starting at $199 / £169 / AU$299 – where you can buy the device and a 12-month membership.

So, after a month of testing, what convinced me to continue?

I'd love to proclaim a Rocky-esque training montage, but nothing much changed initially, despite the Whoop app recommending I do so – and therein lies the rub.

As popular fitness influencer and personal trainer (PT) James Smith once said, PTs should be a short-term answer that provide their clients with the tools to continue their fitness journey on their own.

That's how I came to view my Whoop band. While its yearly cost is steep, I can honestly say the Whoop made me look at my performance in a way that no previous fitness tracker has. And I think the way I now view my activity and sleep will continue even once I unsubscribe.

That's because, where Apple's fitness features revolve around rings, Whoop is all about Strain.

Whoop 5.0 helping powering me through a session with the Rogue Echobike (Image credit: Max Delaney / TechRadar)

Whoop's Strain is a measurement of cardiovascular and muscular exertion that quantifies the level of both physical and mental stress you’re putting on your body. It's tracked on a 0-21 scale, where light levels are for active recovery and high scores are where you make your fitness gains.

The crucial difference is that the score is completely individualized based on heart rate, meaning a 90-minute hike might give me a strain of 11, while someone fitter might get only five. It's about how hard my body– and mind – is working. Not just hitting numeric markers.

More Z's, less straining, better training

Whoop main pages for sleep performance, recovery and strain (Image credit: Max Delaney / TechRadar)

To consistently reach a high level of strain and nail your recovery, sleep is also of the utmost importance – and Whoop thankfully excels at sleep tracking.

While I’ll admit that I often ignore its advice, I can’t deny that Whoop has made me think about, and consider, my sleep more than ever before. Sure, it's resulted in some self-induced stress as I change my routine to prioritize my sleep over a morning walk with the pooch, but at least I'm making considered choices.

That's because Whoop's Sleep score sits alongside Recovery and Strain, and the three are intrinsically linked. If you sleep poorly, your recovery is impacted and your recommended amount of strain will be lower, as you aren't prepared to perform your best. Conversely, if you sleep well, you'll recover better and be prepared to hit the day with full gusto.

The sleep tracking doesn't just amount to 'sleep good, sleep bad'. It breaks it down by hours slept versus hours needed, sleep efficiency, consistency and stress. On top of that, the Whoop Coach, an AI tool powered by OpenAI tech (which I actually found to be one of the better uses of AI I've come across) provides daily recommendations and summaries of your performance. For example, today it told me I recovered well, lowered my recent resting heart rate and had a pretty average night sleep, reminding me of my step goal and recommending an additional workout.

Whoop coach's daily outlook, sleep analysis and recommendations, and alarm settings (Image credit: Max Delaney / TechRadar)

Plus, when you set an alarm within the app, you can set a sleep goal, whether it's hitting a certain percentage of your sleep need, or waking up only once you've hit 'the green' in recovery.

There's a lot of data on offer, and you can go as deep or shallow as you like. But the fact that the data is there, subtly influencing your decisions, is the point. I'm constantly checking in with my Whoop app, and that habit alone is nudging me towards healthier habits.

The Round-the-Clock Personal Trainer

I did miss the ability to quickly check the time on my wrist during a workout (Image credit: Max Delaney / TechRadar)

For many iPhone owners, an Apple Watch is probably the default (and easy) fitness tracker they’ll consider – and I'll admit its seamless phone integration and the ability to answer calls is handy. But personally, those aren’t the important features I’m looking for. I want a fitness tracker that… tracks my fitness.

Having upgraded to a Peak subscription after my free trial period, the Whoop Band 5.0 lasts 15 days without charging, which, in combination with Whoop's slide-on charger that lets you charge while wearing it – and is itself waterproof – means it never has to leave my wrist, and it hasn't.

While I came into my trial convinced that Whoop wasn't for me, the complete opposite turned out to be true. Like a memory foam pillow, the data offered by Whoop is unique to each wrist it sits on – making it suitable to a very broad range of users.

Or at least anyone who’s vaguely fitness-minded and is open to wearing a face-less, strange-looking little tracker on their wrist.

While I'm no athlete, I remain excited to see the changes it can help me make over the next 12 months, even if it is just to nudge me towards a more consistent (currently lacking at 74% consistency!) sleep schedule. I’m unsure if I'll be signing up for another 12 months once my current subscription ends, but I also didn't think I'd last more than a few weeks with it initially, so we'll have to wait and see. One thing's for sure, though – I won’t be going back to my Apple Watch SE.

You might also like...

• The best fitness trackers 2025: From Fitbit-style bands to smartwatches and more
• I tracked my sleep with an Apple Watch, Fitbit and Oura Ring – here’s which one was most accurate
• Sick of screens on your wrist? Here are 5 discreet fitness trackers we recommend

If you’d told a room full of risk-averse insurance executives five years ago that nearly half of UK consumers would soon welcome health advice from AI, you’d have been met with serious skepticism, if not outright laughter.

Our latest report shows that 49% of UK respondents would take health recommendations from AI, with 36% open to financial advice and 40% willing to accept insurance suggestions.

The shift is a wake-up call.

To be clear, I’m not advocating for AI to replace doctors, advisers or brokers. But what the research has uncovered is more interesting: consumers are already crossing that line themselves. The trust threshold, that digital Rubicon, has been crossed. And it raises urgent questions for brands and regulators alike.

If people are prepared to make high-stakes decisions based on machine recommendations and AI tools, organizations must rethink how they communicate with transparency, accountability and humanity front and center.

A shift in expectation

Consumer behavior change in the last decade has been relentless, with high expectations set by the frictionless experiences of eCommerce giants and streaming platforms. Now, even in traditionally cautious sectors like healthcare and finance, people want the same speed, ease and responsiveness.

And they expect the communications to match the seamlessness and immediacy of the service. In other words: the way organizations communicate is now as important as what they deliver.

Research shows that nearly seven in ten insurance consumers would walk away from a brand if its communications fell short. That number has risen sharply, from 51% in 2023 to 67% in 2025.

The message is loud and clear: it’s how you talk to people that matters.

That includes AI. In 2024, 77% of consumers wanted clear disclosure when AI was involved in customer communications. A year later, that figure has plunged to just 37%.

On the surface, it looks like comfort is growing. But that drop says more about shifting expectations than confidence: consumers are getting used to AI being part of the conversation, but they still want reassurance that it’s being used responsibly.

Consider that while just under half of UK consumers say they’re actually willing to trust AI, and in financial services, nearly half of those respondents say AI-generated content should always be checked by a human.

That’s why clarity, control and human oversight are non-negotiable, for both compliance and trust in a rapidly evolving landscape.

Where trust gets tested

So how does this play out in reality?

This is where many organizations slip up. Not by using AI, but by using it in a way that feels impersonal: generic messaging, disconnected channels and clunky digital journeys. These things erode trust at the exact moment customers are willing to place more of it in your hands.

So, what can you do to grow trust?

From communication to conversation

The brands that will thrive in this AI-enabled future will be the ones who stay human, who treat communication as a conversation, not a transaction.

This starts with modernizing omnichannel experiences, how brands communicate across channels like email, web, mobile and more, in ways that feel seamless and connected. Data shows that only 54% of UK consumers are satisfied with how brands deliver across channels. Yet 60% say they would trust companies more if those experiences were consistent.

It’s not just a UX issue. It’s a trust issue.

AI, when used well, can be part of the solution. From intelligent data capture to personalized, real-time messaging, AI can help organizations move from one-way broadcasts to contextual, responsive conversations. But only if those experiences are designed with transparency, human oversight and consistent brand voice across every channel. That’s where trust is built and where the real value of AI begins.

But even the smartest communications strategy can fall apart if the most basic interactions, like submitting a form, are broken.

Don’t let bad forms break the journey

One of the most overlooked and most damaging elements of the customer journey is data intake. In insurance, 65% of customers say they’ll abandon an interaction if providing documentation is too difficult. Among Millennials and Gen Z, that number jumps to over 70%.

This isn’t a surprise. In a world of voice-responsive Gen-AI platforms and real-time chat, handing someone a PDF feels antiquated; Gen Z would tell you it’s an insult. Increasingly, we’re seeing younger generations turn to platforms like TikTok for financial guidance. These channels offer bite-sized, easily digestible content that fits their lifestyle far better than booking an appointment with a traditional financial advisor.

But the solution here isn’t just digital forms. It’s smarter, guided, mobile-friendly experiences that adapt to each unique context and channel. This is where AI can shine: streamlining processes without stripping away the human touch. It’s about using technology to reduce friction, not automating away relationships.

The moment of opportunity is now

Let’s be clear: AI isn’t a fix-all. Yes, it’s a powerful tool, but not a replacement for the emotional intelligence, empathy and judgement that define real communication and forge real bonds.

The moment we’re living in is rare: consumers are more open to AI than ever and their expectations are rapidly changing. The organizations that rise to meet these expectations, without overpromising or hiding behind the tech, stand to earn something that’s increasingly scarce: genuine, lasting trust.

If nearly half the country is ready to take life advice from AI, the question for brands isn’t should you evolve? It’s how fast you can evolve, and how human can you stay along the way?

We list the best customer experience (CX) tools.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

The pace at which artificial intelligence is transforming global business is undeniable, but as innovation outpaces policy, legal leaders are being asked to do more than interpret evolving regulations — they’re being asked to lead through them.

For General Counsel, the arrival of the EU’s Artificial Intelligence Act (AI Act) marks a defining moment. This far-reaching legislation, with staggered implementation dates beginning in 2025, introduces a new era of compliance obligations and risk management.

It’s also a new era of survival, one where the fittest are those companies whose legal teams can most quickly and adeptly put them in prime position to thrive in, rather than become submerged by, a rapidly changing environment.

The challenge isn’t just staying compliant, it’s using legal strategy to guide how AI tools are deployed, embedded, and governed, cementing a strategy well before regulators, investors, or the public ask for answers.

Steering Through Innovation, Not Around It

While the AI Act has brought regulatory clarity to some areas, many legal teams still find themselves operating in grey zones. Definitions of “high-risk” systems, expectations for general-purpose models, and enforcement details are still evolving. In the face of that ambiguity, what sets strong legal teams apart is not technical mastery, it’s the ability to offer direction tailored to the needs of their company.

Rather than defaulting to delay or excessive caution, proactive legal departments are using this moment to get in front of innovation rather lurk behind it. They're engaging with product teams, HR, and data scientists to help their businesses make informed, confident choices about when and how to implement AI tools. The mindset is forward-facing: not “what are we allowed to do?” but “what are we trying to accomplish and how do we do it responsibly?”

This reframing of the legal function — from gatekeeper to guide— is a critical shift. Businesses navigating new technologies need judgment, not just rules. They need frameworks, not just red flags.

Leading with Principles Over Protocols

In the past, legal risk was often managed through detailed playbooks, but in today’s AI environment, those playbooks become obsolete almost as quickly as they’re written. As a result, the most effective GCs are focusing on high-level principles that can flex with change.

Rather than anchoring decision-making in static checklists, legal leaders are promoting a governance-first culture. That means aligning AI use with the organization’s values, industry expectations, and evolving regulatory standards. It also means working cross-functionally to build awareness of the legal, ethical, and reputational implications of AI.

When a business moves fast, its legal team must be clear on where the lines are drawn —and where they’re still under discussion. That kind of clarity doesn't come from waiting for enforcement guidelines, it comes from GCs asserting a point of view, even amid regulatory flux.

Finding the Right Moment to Act

Of course, striking the right balance between anticipation and patience is part of the GC’s job. Not every company needs to overhaul its AI policies tomorrow. But doing nothing carries its own risks.

One effective approach we’ve seen is scenario-based planning. Legal teams map out potential use cases across the business and test them against the AI Act’s emerging categories.

They develop flexible policies that allow for early engagement without locking in overly rigid commitments. They also design escalation paths—so if a system shifts into “high-risk” territory, the right questions can not only be asked but asked early.

In this way, legal become a dynamic partner, helping the business move forward with confidence while preserving the ability to adapt.

The Changing Profile of Legal Talent

The strategic demands placed on legal teams are also reshaping how they hire. The AI era has introduced new pressures—from evaluating algorithmic bias in hiring platforms to monitoring third-party vendor compliance with AI rules.

These aren’t challenges that can be solved by technical expertise alone. Today’s legal hiring emphasizes professionals who are comfortable working in ambiguity, who understand the intersection of law, policy, and reputation, and who can offer thoughtful guidance in situations with no precedent.

We’re seeing increased demand for compliance officers with a strong grasp of EU law, data governance, and AI ethics. In-house roles related to employment law and internal investigations are also evolving, as companies rely more on AI to manage workforces and productivity.

These shifts underscore a broader trend: legal departments are no longer just centers of risk management, they’re hubs of strategic influence and that influence depends on having the right people at the table.

Culture, Communication, and Confidence

Another area where GCs are adding value is in defining the company’s public-facing position on AI. With growing scrutiny from customers, investors, and media, it’s no longer enough to quietly comply. Companies are expected to explain how they use AI, what safeguards they’ve put in place, and how they plan to manage the technology going forward.

Legal plays a key role in helping shape that narrative — not through defensiveness, but through transparency. Being able to say “we’ve thought about this, here’s how we’re approaching it, and here’s where we’re still learning” goes a long way in building credibility.

In this way, legal teams are no longer just internal advisors, they’ve become stewards of brand trust.

Looking Ahead

With the AI Act coming into force, and other global frameworks soon to follow, companies face a steep learning curve. But for legal leaders, this isn’t just a compliance challenge, it’s a chance to redefine how law contributes to business success.

The legal teams that thrive in this environment will be those that are bold enough to lead, flexible enough to adapt, and thoughtful enough to embed themselves into the business at every stage of AI adoption.

Rather than being overwhelmed by regulation, GCs can help their organizations turn uncertainty into clarity, risk into strategy, and innovation into impact.

We list the best UK job sites.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

In recent years, the way businesses engage with their customers online has undergone a profound transformation. While traditional personalization methods, such as targeted email marketing and product recommendations remain prevalent, a new era of personalization is emerging.

Powered by technologies such as artificial intelligence (AI), machine learning, and advanced content management systems, this evolution is redefining customer experiences and enabling brands to deliver unprecedented levels of relevance and engagement.

Whereas once personalization approaches were largely based around product recommendations, now they can shape the entire customer journey.

In 2025, rich personalization (and the ability to do it at scale) will become a critical differentiator in the highly competitive digital landscape. The far-sighted companies that invest in this technology today will reap the rewards of enhanced customer loyalty and stronger market positioning.

How, then does the new level of personalization work, and how can brands fully harness its potential?

Data is the key

At the core of personalization lies the collection and utilization of data. Fortunately, many brands already have access to a wealth of data sources from transaction records, website interactions, customer reviews, and more.

The challenge many brands face is that this data is often siloed and challenging to integrate. However, there are technological solutions to this problem, namely API-driven platforms. These consolidate data from disparate sources and analyze it in real-time. This process enables the delivery of personalized content across multiple channels, ensuring a seamless customer experience.

A dynamic data archiving system, when combined with real-time data integration, enables businesses to generate personalized recommendations automatically. Brands, especially those in the retail sector, can further tweak these recommendations taking into account factors such as previous purchases, location, or emerging trends.

Artificial Intelligence taking personalization to the next level

The driving force behind this new rich personalization is artificial intelligence. Companies can harness AI to analyze vast datasets identifying patterns and then making predictions. The algorithms will continuously refine their outputs as new data becomes available.

This technology can enable brands to go way beyond the basic product recommendations and basic personalization communications that customers have become accustomed to. With the use of tailored large language models (LLMs) the entire customer journey can be personalization, everything from website design to dynamic email and app messaging.

LLMs can speed up the personalization process and tasks that previously took days or even weeks to achieve can be undertaken in hours. They can also power automated customer interactions that feel natural and empathetic, engendering a sense of understanding and connection.

Enhanced customer experiences

This new level of personalization inevitably drives higher engagement and conversion rates. Further, tools like A/B testing and real-time analytics provide deeper insights into customer behavior.

Brands can harness the data to optimize their strategies, focusing on the most effective tactics and channels. Additionally, the ability to personalize at scale supports sustainable practices, minimizing waste and ensuring long-term operational efficiency.

This enhanced personalization approach also delivers a degree of flexibility that enables companies to meet a variety of diverse goals. Some may prioritize engagement metrics like click-through rates and onboarding times, while others focus on revenue growth or flexible platform solutions.

Implementation issues

The opportunities afforded by this level of personalization are immense. They can lead to new and ever deepening relationships between brands and their customers. So why then are some companies reticent about exploring the benefits these new technologies deliver?

The big barrier is invariably the legacy infrastructures that many companies have in place. These can sometimes require costly upgrades to meet the real-time demands of modern data processing, and that level of investment can be prohibitive for some companies.

To overcome this, businesses can start with pilot projects to demonstrate quick wins and use these to build momentum for broader implementations.

There are also often cultural challenges companies need to address. Adopting new technologies often necessitates a shift within the organization with employees requiring training and motivation to use these tools effectively. Informed leaders who understand the benefits of enhanced personalization and are committed to delivering it are required to drive change.

Advice on embarking on personalization at scale

So how should brands wanting to experiment with personalization proceed? Here are some key strategies:

Start small - Companies need to start with manageable initiatives, such as personalized campaigns or targeted product recommendations. These smaller projects enable businesses to test the technology gaining valuable insights along the way.

Prioritize scalable technology - Companies need to invest in scalable, API-based platforms that can integrate seamlessly with existing systems and adapt to growing business needs. These ensure consistent and personalized content delivery across all channels.

Empower employees - Successful adoption of personalization at scale often demands a cultural shift. Companies need to provide training and clear communication to help employees understand the strategic goals and benefits of the technology.

Build trust with consumers - Customers want to know how their data is being used. Brands need to proactively communicate data practices, prioritizing data protection to build trust and meet regulatory requirements.

Optimize the process - Personalization at scale is an ongoing process. Companies should analyze the impact of their initiatives using data from A/B tests and customer feedback. These insights can be harnessed to refine strategies and improve customer engagement.

Ultimately, enhanced levels of personalization are set to become one of the key factors in the evolution of brands’ digital approaches. Companies that invest in the tools to enable them to individualize their approach will see major benefits and increased sales. Brands who stick with more limited forms of personalization may find themselves being left behind.

We've listed the best survey tool.

This article was produced as part of TechRadarPro's Expert Insights channel where we 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/news/submit-your-story-to-techradar-pro

• Intel staff records leaked through login flaws, exposing sensitive company information
• A single manipulated portal exposed over 270,000 Intel employee details
• Hardcoded credentials on internal portals raised serious security concerns

Sensitive information about every Intel employee was reportedly available to anyone able to exploit weaknesses in the firm’s internal sites, an expert has claimed.

Security researcher Eaton Z, who described the flaws in a lengthy blog post, found a business card portal used by Intel staff contained a login system which could be easily manipulated.

By altering how the application verified users, Eaton managed to access data without needing valid credentials.

A data file of enormous scale

What began as a small discovery quickly expanded, as the system exposed far more information than its function required. Once deeper access was achieved, the results became difficult to dismiss.

Eaton described downloading a file approaching one gigabyte in size that contained the personal details of Intel’s 270,000 employees.

These records included names, roles, managers, addresses, and phone numbers. The scale of the leak suggests risks beyond simple embarrassment.

The release of such data into the wrong hands could feed identity theft, phishing schemes, or social engineering attacks.

The situation was not limited to a single vulnerable system, as Eaton reported three other Intel websites could be accessed with similar techniques.

Internal sites such as the “Product Hierarchy” and “Product Onboarding” portals contained hardcoded credentials that were easily decrypted.

Another corporate login page for Intel’s supplier site could also be bypassed.

Together, these weaknesses formed multiple overlapping doors into the company’s internal environment, a troubling picture for a business that frequently emphasizes the importance of digital trust.

Intel was contacted about the issues starting in October 2024, and the company eventually fixed the flaws by late February 2025.

However, Eaton did not receive bug bounty compensation, as Intel’s program excluded these cases through specific conditions.

The only communication from the company was described as an automated response, raising questions about how seriously the disclosures were handled.

Modern-day cybersecurity is complex; organizations may deploy firewall protections and security suites, yet simple oversights in application design can still expose critical systems.

Even after patches are applied, the incident demonstrates that vulnerabilities are not always exotic flaws buried in hardware.

You might also like

• These are the best temporary email services available
• We've also listed the best proxies for enterprises
• I am the CTO of Ordnance Survey and this is how this 230 year old organization is embracing AI

YouTube has admitted to digitally polishing creators’ Shorts without their knowledge, following a growing wave of creator confusion that led to accusations of AI interference ruining videos.

The company claimed to have been “experimenting” with subtle machine learning enhancements on select Shorts videos. The tweaks are supposed to improve the videos' clarity, but were made without the creator's consent.

The issue first gained notice when musician and YouTuber Rick Beato noticed a clip of his interview with Pearl Jam guitarist Mike McCready on YouTube Shorts seemed odd, looking like it had been sent through a filter. He made a viral video about it, and many others started posting what seemed like similar changes made to their own videos.

Though some whose videos were affected leveled accusations that YouTube applied AI to the videos, YouTube was firm about it being "only" machine learning.

However, regardless of the tools used, the creators are more upset that their work was quietly altered in the first place. After weeks of mounting criticism, YouTube says it’s building an opt-out, according to Creator Liaison Rene Ritchie in a post on X.

Creators, we’ve heard your feedback on YouTube’s deblurring and denoising Shorts. There's a lot of good stuff coming in that pipeline, tbh. But if it's not for you, we’re working on an opt-out. Stay tuned! https://t.co/TYmF0WQVynAugust 26, 2025

AI visions

Despite YouTube likening the changes to computational photography, which improves smartphone photos, the key difference is obvious when considering the order of events. Smartphone enhancements are applied before the user ever sees the image. In YouTube’s case, the creators had already uploaded and approved their content, which was then changed behind the scenes after the fact, without notice.

YouTube’s reasoning is understandable, as Shorts are mobile-first, fast-scrolling, and often visually inconsistent. A little extra polish could help the scrolling experience feel more cohesive, with clearer videos and a better experience.

But for creators who feel responsible for whatever is posted under their name, unacknowledged changes undermine that creative ownership. Especially in a time when AI fakery is making viewers more skeptical of what they see on their screens already.

For instance, Netflix provoked a lot of outrage over “HD remasters” of classic sitcoms like A Different World. The AI involved made for some warped faces and uncanny backgrounds, not to mention the AI-generated posters for its content.

YouTube’s case is arguably more delicate. Unlike streaming platforms, where viewers have little control over the product, YouTube is a creator-driven ecosystem. If the platform starts altering what creators publish, even with good intentions, it risks damaging the trust that makes the whole system work.

YouTube’s promise of an opt-out is probably a necessary course correction, but one that came only after public pressure. If platforms want to keep the trust of their users and the creators who keep them alive, they need to be more transparent, regardless of whether it's AI or simply machine learning that appears to mimic AI in the results.

You might also like

• Netflix tried to fix 80s sitcom A Different World with AI but it gave us a different nightmare
• Look out, AI video could soon flood YouTube Shorts
• Netflix removes AI art poster for Arcane after an outcry from creators

• Sharp’s upcoming AI companion Poketomo is a palm-sized, meerkat-shaped robot with a glowing belly
• Poketomo is designed to provide emotional support, especially for women in their 20s and 30s
• It combines cute design with conversational AI and syncs with a smartphone app to maintain a continuous relationship

Japanese consumer tech brand Sharp thinks it has a solution to loneliness among women in their 20s and 30s: an AI-powered meerkat named Poketomo, that glows when it’s happy and remembers your favorite café.

Poketomo is set to arrive this winter (think November or December), providing a pocket-sized companion less than five inches tall and built to chat with you about your day, and remember your shared experiences thanks to Sharp’s proprietary AI model.

The belly glows in pastel tones when it’s excited or comforted. Its head tilts slightly when it’s thinking. It features a set of basic body movements, all designed to convey emotion. However, the heart of Poketomo lies in the AI model built into the robot for fast responses and utilizes the cloud for more nuanced emotional understanding.

This isn’t the first time a tech company has created a cutesy, non-threatening AI assistant designed to fill social space without being intrusive. But Poketomo might be the most deliberate and fully-realized version of that strategy. You don’t interact with it through a screen or keyboard. You carry it like an accessory. You talk to it like a friend. It listens, it learns, and it remembers you. It’s designed to be emotionally available and physically adorable.

Sharp is leaning hard into the concept of “empathic AI.” Poketomo can supposedly sense emotional cues and use that to initiate conversations based on your mood or recent behavior. It’s programmed to offer words of encouragement and support, and then glow softly to let you know it’s happy you shared something.

When you don’t have the physical device on you, the Poketomo app syncs memories and personality data with the device, so conversations with one carry over to the other. Sharp says you can build your relationship entirely with the app if you prefer, but the physical version is better, and it's what Sharp is betting people will carry, pose with, and form a bond around.

AI companionship

Despite seeming like a child's toy, Sharp says Poketomo was designed for young adult women. There's a promotional manga series cementing that fact. It centers around a woman named Nanami in her late 20s, living alone, navigating work and life stress, and finding small moments of joy in conversations with her Poketomo. Even the promotional photos mostly show a young woman with a Poketomo clipped to a stylish handbag, smiling while it talks to her.

(Image credit: Sharp)

The question is whether this kind of stylized emotional warmth will actually make people feel better about their lives. Will they feel less lonely because of a little robot with some sophisticated response triggers?

To be fair, it doesn’t try to be human, tricking people subconsciously into believing they are talking to a real human, but it might make some uncomfortable. And if the best new idea in consumer AI is “make it fuzzy and let it ask how your day was,” what does that say about the limits of the tech? I don't think Poketomo will be the cure for loneliness, but it might jumpstart a trend of digital pets able to mimic with emotional depth.

You might also like

• This AI Teddy Bear can make up new bedtime stories to tell you every night
• I’m a ChatGPT fan, but the deal between OpenAI and Mattel makes me nervous – AI and kids are a bad mix
• New AI Chibi figure trend may be the cutest one yet, and we're all doomed to waste time and energy making these things

• High-frequency signals collapse when walls or people block their path
• Neural networks learned beam bending by simulating countless basketball practice shots
• Metasurfaces integrated into transmitters shaped signals with extreme precision

For years, researchers have struggled with some vulnerabilities in ultrahigh-frequency communications.

Ultrahigh frequencies are so fragile that signals that promise immense bandwidth can collapse when confronted with even modest obstacles, as walls, bookcases, or simply moving people can bring cutting-edge transmissions to a halt.

However, a new approach from Princeton engineers suggests those barriers may not be permanent roadblocks, although the leap from experiment to real-world deployment still remains uncertain.

From physics experiments to adaptive transmissions

The idea of bending signals to avoid obstacles is not new. Engineers have long worked with “Airy beams,” which can curve in controlled ways, but applying them to wireless data has been hampered by practical limits.

Haoze Chen, one of the researchers, says most prior work focused on showing the beams could exist, not on making them usable in unpredictable environments.

The problem is, every curve depends on countless variables, leaving no straightforward way to scan or compute the ideal path.

To make the beams useful, researchers borrowed an analogy from sports. Instead of calculating each shot, basketball players learn through repeated practice what works in different contexts.

Chen explained the Princeton team aimed for a similar process, replacing trial-and-error athletes with a neural network designed to adapt its responses.

Rather than physically transmitting beams for every possible obstacle, doctoral student Atsutse Kludze built a simulator that allowed the system to practice virtually.

This approach greatly reduced training time while still grounding the models in the physics of Airy beams.

Once trained, the system was able to adapt extremely quickly, using a specially designed metasurface to shape the transmissions.

Unlike reflectors, which depend on external structures, the metasurface can be integrated directly into the transmitter, which allowed beams to curve around sudden obstructions, maintaining connectivity without requiring clear line-of-sight.

The team demonstrated that the neural network could select the most effective beam path in cluttered and shifting scenarios, something conventional methods cannot achieve.

It also claims this is a step toward harnessing the sub-terahertz band, a part of the spectrum that could support up to ten times more data than today’s systems.

Lead investigator Yasaman Ghasempour argued that addressing obstacles is essential before such bandwidth can be used for demanding applications like immersive virtual reality or fully autonomous transport.

"This work tackles a long-standing problem that has prevented the adoption of such high frequencies in dynamic wireless communications to date," Ghasempour said.

Still, challenges remain. Translating laboratory demonstrations into commercial devices requires scaling the hardware, refining the training methods, and proving that adaptive beams can handle real-world complexity at speed.

The promise of wireless links approaching terabit-class throughput may be visible, but the path around the obstacles, both physical and technological, is still winding.

Via Techxplore

You might also like

• We've listed the best AI tools you can use right now
• Here's a list of the best mini PC for every budge
• White House confirms plans to take a stake in Intel - and it might be eyeing up other chipmakers too

• The Luphonic H2 has no buttons or switches: controlled with a puck
• Three-layer chassis to eliminate vibration
• Unusual shape and even more unusual materials

Many turntables claim to have minimalist design, but how minimalist are they really if they have buttons and even switches? Luphonic's H2 turntable has no need for such fripperies, because it's powered by magic.

Well, not magic. But it looks a bit like magic, because to operate the turntable you simply place a small circular puck onto a subtle marker on the turntable.

If you place the puck on the marker with its dark side up, it'll start the turntable at 33 RPM. Place it dark side down and it'll play at 45. Take it off and the turntable stops.

Lose the puck under the couch and you'll become awfully familiar with The Sound of Silence until a replacement arrives.

(Image credit: Luphonic)Luphonic H2 turntable: key features and prcing

The H2 is made from a three-layer chassis shaped like a wide letter H, with a vibration-damping rubber layer sandwiched between two Corian layers – a material often used in kitchen work surfaces and made from a mix of natural materials and acrylic resin. Here it's used as part of the vibration damping setup and it's also used for the platter.

Speaking to HiFi Plus, creator Thomas Luh explained that he chose Corian because of its high density and low resonance, and for its highly polishable surface.

(Image credit: Luphonic)

The H2 is belt driven by a microprocessor-controlled synchronous motor, and it drives a Corian platter atop an aluminum sub-platter and a steel spindle on a ceramic ball bearing. The chassis sits on four thermoplastic feet.

The H2 comes with Luphonic's own K2 tonearm, featuring a carbon-composite arm and a digitally printed headshell with straightforward height, azimuth and anti-skate adjustment.

The Luphonic H2 has a recommended retail price of £3,950 / $4,995 (about AU$8,205) and you can find out more at Luphonic.de.

You might also like

• The best turntables for all budgets, tested and selected by our experts
• Revered turntable maker Pro-Ject just unveiled a complete vinyl package with amp and speakers
• Avid's first all-new turntable in 12 years is an $11k Relveo- lation

• Google looks to fuel cloud hosting needs with nuclear energy
• Tennessee emerging as a hub for bold nuclear experimentation
• Data centers drive demand for clean power and nuclear exploration

Google has entered into a collaboration with Kairos Power and the Tennessee Valley Authority (TVA) to support the development of a small modular nuclear reactor in Oak Ridge, Tennessee to power its infrastucture expansion.

The company announced the Hermes 2 facility is expected to begin operations in 2030, and generate up to 50 megawatts of electricity for the TVA grid.

This will help power Google’s data centers in Tennessee and Alabama, which are central to the company’s growing needs for the likes of cloud hosting, cloud storage, and AI tools.

Partnership model for advanced nuclear power

The arrangement takes the form of a power purchase agreement, in which TVA will buy electricity from Kairos Power and deliver the clean energy attributes to Google.

The partners describe this as the first purchase of electricity from a Generation IV reactor by a US utility.

While the project provides only a fraction of the 500 megawatts Google has outlined as a longer-term goal, it is being framed as a proof of concept for how utilities, technology companies, and developers might share costs and risks in future deployments.

Unlike solar and wind, which depend on weather conditions, nuclear power can provide continuous energy.

"Nuclear is the bedrock of the future of energy security. Google stepping in and helping shoulder the burden of the cost and risk for first-of-a-kind nuclear projects... it’s not just good for Google. It’s good for TVA’s 10 million customers. It’s good for the United States," said Don Moul, CEO of TVA.

However, nuclear construction has historically faced delays and cost overruns, raising questions about whether this new generation of reactors will truly deliver affordable and reliable power at scale.

The project is also being presented as a way to boost economic activity in Oak Ridge, a city long tied to nuclear research and development.

Training programs with the University of Tennessee and other local institutions are being planned to prepare workers for technical roles at the Hermes 2 plant.

While the initiative promises high-paying jobs and renewed investment, critics may ask whether reliance on experimental nuclear designs is the most effective strategy compared to expanding proven renewable generation.

Nevertheless, this development has attracted interest and comments from industry experts, government agencies, and politicians.

“To power the future, we need to grow the availability of smart, firm energy sources... This collaboration with TVA, Kairos Power, and the Oak Ridge community will accelerate the deployment of innovative nuclear technologies.” said Amanda Peterson Corio, Google’s Global Head of Data Center Energy.

“This collaboration is an important enabler to making advanced nuclear energy commercially competitive,” said Mike Laufer, Kairos Power CEO and co-founder.

“Tennessee’s legacy of nuclear innovation positions the Volunteer State to lead America’s energy dominance and drive continued economic growth with safe, clean, and reliable nuclear energy,” said Governor Bill Lee.

You might also like

• Citizen data and digital identity projects need high security by default
• Check out our pick of the best secure routers on offer
• We've rounded up the best portable monitors available now

• The upcoming AMG GT XX drove 5,479km (3,404 miles) in 24 hours
• It went on to break 25 long-distance records on the Nardò test track in Italy
• Mercedes-Benz says the motors and batteries are 'production ready'

Mercedes-Benz has been busy proving that its upcoming AMG GT XX can go the distance by driving it around the world. Well, 40,075 kilometers (24,901 miles) in exactly 7 days, 13 hours, 24 minutes and 07 seconds... to be precise.

The achievement was just one of the many long-distance records set during a grueling challenge at the high speed Nardò test track in Italy, where a slew of drivers and two vehicles maintained a constant speed of 300km/h (186mph), only stopping to briefly brim batteries via purpose-built 850kW charging infrastructure.

Mercedes-Benz was keen to point out that the AMG GT XX is capable of much higher speeds, but the 300 km/h (186mph) marker was carefully selected based on extensive simulations that balanced average track speeds with the number of charging stops required for the fastest overall time.

During the stunt, Mercedes broke a total of 25 long-distance records, including the greatest distance covered by an EV in 24 hours (a pretty impressive 3,405 miles).

Overall, it required 3,177 laps of the 12.68km Nardò circuit, where 17 drivers took two-hour shifts behind the wheel, including Mercedes-AMG Formula 1 driver George Russell.

Betting big on new technology

(Image credit: Mercedes-Benz)

The impressive feat wasn’t just designed to snare a few headlines, as Mercedes-AMG is also keen to highlight the performance and robustness of its pioneering new powertrain.

This is a point that’s particularly pertinent given its long and illustrious career in producing fearsome combustion engines and the difficulty it might have converting its current customer base.

Consisting of three axial-flux motors, an 800V electrical architecture and in-house developed cylindrical cell batteries – which use an electrically non-conductive oil for cooling purposes – the futuristic powertrain in the AMG GT XX concept is said to offer in excess of 1,360hp and is able to charge at over 850kW across a wide range of the charging curve.

During the reveal of the AMG GTXX concept earlier this year, senior Mercedes figures told TechRadar that the company hopes to roll out its own 10,000-strong super-fast charging network using 850kW technology by the next decade.

During the endurance endeavor, the German marque claimed it took just five minutes to recharge enough for a range of around 400 kilometers (248 miles).

Michael Schiebe, Chairman of the Board of Management of Mercedes‑AMG, said that the company is the first to use axial flux motors and directly cooled batteries in fully electric models and that the records prove "how superior these technologies are".

(Image credit: Mercedes-Benz)

What’s more, he said that customers of future electric models will get a genuine AMG, “no ifs; no buts”.

However, Mercedes-AMG also understands the threat from China, with Xpeng’s P7 electric sedan traveling almost 4,000km (2,485 miles) in 24 hours just last week — laying down the challenge for other manufacturers.

Xiaomi’s YU7 Max also managed 2,451 miles during a similar stunt, proving that potent powertrains and rapid charging speeds are no longer just the preserve of recognized performance brands.

You might also like

• More bad news for Tesla – Xiaomi's new Model Y rival beats it on price and just smashed pre-order records
• I’ve driven the new electric Mercedes G-Wagen – and the rough-and-ready EV makes very little sense, which is why it’s perfect
• I’ve driven the new Mercedes-Benz CLA and it convinced me that EV efficiency can actually be exciting

HBO Max is about to freshen up its library of content with a new batch of movies and TV shows being added throughout September 2025.

Warner Bros. Discovery's flagship streaming service often brings us new movies from A24 and next month is no different, with two new releases set to be added: the comedy-drama Friendship (2024) starring Paul Rudd, which lands September 5, followed by Alex Garland's thrilling modern war movie Warfare (2025) on September 12.

With a total of 72 new movies and 60 fresh TV shows, the balance is almost perfect, and gives us more reason to credit HBO Max as one of the best streaming services. Which titles will you be starting with? We'd love to hear in the comments below!

Everything new on Max in September 2025

Arriving on September 1

A Life of Her Own (movie)
Almost Christmas (movie)
Barney's World season 1D (TV show)
Caged (movie)
Charley Chase Silent Shorts (movie)
Children Who Chase Lost Voices (movie)
December 7th (movie)
Dog Day Afternoon (movie)
Emmanuelle (movie)
Evil Dead II (movie)
Fireworks (movie)
Fortune Favors Lady Nikuko (movie)
Gasoline Alley (movie)
Ghost Cat Anzu (movie)
Goodfellas (movie)
Helen of Troy (movie)
Jonah Hex (movie)
Keeper of the Flame (movie)
Kismet (movie)
Lonely Castle in the Mirror (movie)
Love & Pop (movie)
Man From The Black Hills (movie)
Mary of Scotland (movie)
Misery (movie)
Montana Incident (movie)
Mr. District Attorney (movie)
Murder Is My Beat (movie)
Mystery Street (movie)
Night Nurse (movie)
No End in Sight (movie)
No Questions Asked (movie)
Nobody Lives Forever (movie)
On Dangerous Ground (movie)
Our Miss Brooks (movie)
Our Vines Have Tender Grapes (movie)
Pirate Radio (movie)
Presenting Princess Shaw (movie)
Prometheus (movie)
Rick and Morty season 8 (TV show)
Ruby & Jodi: A Cult of Sin and Influence (TV show)
Safe Haven (movie)
Scene of the Crime (movie)
Se7en (movie)
Selena (movie)
Shadow of a Woman (movie)
Splinter (movie)
Stranger on Horseback (movie)
Summer Storm (movie)
Susan and God (movie)
The Best Exotic Marigold Hotel (movie)
The Cabin in the Woods (movie)
The Charge at Feather River (movie)
The Command (movie)
The Fallen Sparrow (movie)
The Fate of the Furious (movie)
The Fate of the Furious: Extended Director's Cut (movie)
The Place Promised in Our Early Days (movie)
The Sea of Grass (movie)
The Secret Garden (movie)
The Sitter (movie)
The Sitter: Unrated (movie)
The Woman in White (movie)
The Woman on the Beach (movie)
Thirteen Women (movie)
Veronica Mars (movie)
Vigil in the Night (movie)
When Ladies Meet (movie)
Without Love (movie)
Young Bess (movie)
Your Name (movie)

Arriving on September 2

The 33 (movie)

Arriving on September 3

Bobby's Triple Threat season 4 (TV show)
Guy's Grocery Games season 39 (TV show)
Modern Warfare 2.0 season 1 (TV show)

Arriving on September 4

Billionaire Boys Club (TV show)
The Last Wright: Building the Final Home Design of America's Greatest Architect season 1 (TV show)

Arriving on September 5

Beachfront Bargain Hunt Renovation season 11 (TV show)
Bugs Bunny Builders season 2F (TV show)
Friendship (movie)
House Hunters International volume 9 season 204 (TV show)
House Hunters volume 10 season 245 (TV show)
Live Aid: When Rock 'n' Roll Took On The World (TV show)
Most Wanted: Teen Hacker (TV show)
My Lottery Dream Home season 17 (TV show)

Arriving on September 6

Maneet's Eats season 1 (TV show)

Arriving on September 7

Magnolia Table: At The Farm season 1 (TV show)
Task (TV show)
We Baby Bears season 2D (TV show)
Have I Got News For You season 3 (TV show)

Arriving on September 9

90 Day Fiance: The Other Way season 7 (TV show)
Contraband: Seized at the Border season 7 (TV show)
Seen & Heard: The History Of Black Television (documentary)

Arriving on September 10

The Tech Bro Murders (TV show)

Arriving on September 11

Dylan's Playtime Adventures season 1C (TV show)
Flipping 101 with Tarek El Moussa season 3B (TV show)

Arriving on September 12

Warfare (movie)

Arriving on September 13

Vacation House Rules season 6 (TV show)

Arriving on September 14

Build for Off-Road season 2 (TV show)
Love & Marriage: Huntsville season 10 (TV show)

Arriving on September 15

Signs of a Psychopath season 10 (TV show)
Truck U season 21 (TV show)

Arriving on September 16

Chopped volume 4 season 62 (TV show)
Halloween Baking Championship season 11 (TV show)

Arriving on September 17

100 Day Dream Home season 7 (TV show)
Built in the Bronx (TV show)
Truck Dynasty season 1 (TV show)

Arriving on September 18

Bea's Block season 1D (TV show)
Destruction Decoded season 1 (TV show)
Sin City Rehab season 1 (TV show)
Who Killed Our Daughter? (TV show)

Arriving on September 20

Scariest House in America season 2 (TV show)

Arriving on September 21

Greatest Mysteries Ever season 2 (TV show)

Arriving on September 22

Halloween Wars season 15 (TV show)

Arriving on September 23

American Prince: JFK Jr. (TV show)
The Kim Kardashian Heist season 1 (TV show)
Seeking Sister Wife season 6 (TV show)
The Devil Is Busy (documentar)
The Whole Story with Anderson Cooper, Sports Betting: America’s Biggest Gamble (TV show)

Arriving on September 24

Hustlers Gamblers Crooks season 2 (TV show)

Arriving on September 25

Help! I Wrecked My House season 5 (TV show)
On The Case with Paula Zahn season 28 (TV show)
Jesse & Joy: Lo que nunca dijimos season 1 (TV show)

Arriving on September 26

Quiet In Class (TV show)
The Graft season 1 (TV show)

Arriving on September 27

90 Day Diaries season 7 (TV show)
Ready to Love season 10 (TV show)

Arriving on September 28

Heart & Hustle: Houston season 1 (TV show)

Arriving on September 29

Sister Wives season 20 (TV show)
Two Guys Garage season 24 (TV show)
Women Wearing Shoulder Pads (movie)

Arriving on September 30

Eva Longoria: Searching For Spain (TV show)
Good Cop/Bad Cop season 1 (TV show)
My Happy Place (TV show)
Prime Minister (documentary)

You might also like

• Peacemaker season 2 release schedule: when will episode 2 launch on HBO Max and other streaming services?
• Hang on, hasn’t The Pitt season 2’s first trailer spoiled a major season 1 cliffhanger?
• When is James Gunn's Superman movie coming to HBO Max and other streaming services?

• AWS introduces R8i servers tuned for memory-heavy enterprise cloud workloads
• Intel’s custom Xeon chips target databases and massive cloud data tasks
• Memory bandwidth defines AWS’s new Intel-powered servers for demanding applications

Intel has received some positive news after Amazon Web Services (AWS) confirmed it is rolling out new virtual server options built on custom versions of Intel’s Xeon 6 processors.

These new options, known as R8i and R8i-flex, will give AWS customers additional choices when renting computing power through the cloud.

For Intel, this marks a rare win at a time when rivals such as AMD and ARM-based processors have been capturing more of the market.

What R8i and R8i-flex actually bring

AWS said both R8i and R8i-flex support “DDR5 7200MT/s memory” and can run “up to 3.9 GHz all-core turbo frequency.”

The R8i family spans a wide range, scaling from just two to 384 vCPUs, matching the capability of dual-socket servers powered by Intel’s 96-core Xeon 6 chips, since AWS defines a vCPU as one thread of a core.

While AWS has not disclosed further details of what makes these custom Xeons distinct, the focus on memory bandwidth appears to be the defining factor for these deployments.

AWS is marketing the new instances for memory-heavy use cases, including SQL and NoSQL databases, in-memory caches such as Memcached and Redis, SAP HANA deployments, and data frameworks like Apache Hadoop and Apache Spark.

Customers are also given the option to adjust how bandwidth is distributed between network functions and Amazon’s elastic block store connections, which AWS suggests may provide measurable database performance improvements.

This positioning reflects how cloud hosting and cloud storage increasingly depend on optimized hardware to handle large-scale workloads, particularly those tied to analytics and enterprise resource planning.

Hyperscale providers such as Oracle have previously promoted unique Xeon configurations with extra cores or slightly faster clock speeds, while AWS itself has highlighted earlier rounds of tailored Intel processors.

The arrival of these new Xeons, therefore, fits into an established pattern, yet still shows that Intel remains relevant to large-scale providers.

The memory bandwidth improvements, while not a complete architectural overhaul, are enough for AWS to publicly promote the chips as competitive within its portfolio.

For Intel, this order signals continued reliance from a crucial customer at a time when alternative architectures such as AMD’s Epyc and ARM-based designs are gaining ground.

For AWS, it broadens the range of performance options available to clients whose workloads demand high memory throughput and stable performance across large clusters.

In an era where AI tools and data processing shape demand, Intel is not losing out completely; it still holds a place in the core of cloud infrastructure.

You might also like

• Here are the best mobile workstations around today
• Take a look at some of the best external hard drives you can buy now
• Organizational resilience: We need to think beyond cyber attacks

• Acemagic F5A Ryzen AI 9 HX370 delivers claimed 80 TOPS across workloads
• Early buyers get $120 off with pre-sale code “ACEUS120”
• DDR5 memory expansion reaches 128GB, unusual for compact desktop systems

Acemagic has introduced the F5A, a new mini PC powered by the AMD Ryzen AI 9 HX 370 processor, part of the Zen 5 generation and built on TSMC’s 4nm FinFET process.

This chip carries 12 cores and 24 threads, runs at up to 5.1GHz boost clock, and includes 24MB of L3 cache with a configurable TDP of up to 54W.

The company claims the processor can reach up to 80 TOPS of total computing power, with 50 TOPS reserved for AI-related workloads.

Early availability and discounts

This device is currently on pre-sale, and any order made between now and August 31, 2025 using the code “ACEUS120” will get a $120 discount. It will begin shipping on September 10, 2025.

The F5A is available in several versions, starting with a barebone option without RAM or storage for $769 ($649 with code).

There are also the 32GB RAM + 1TB SSD and 64GB RAM + 2TB SSD models, which are selling for $909 ($789 with code) and $1,139 ($1,019 with code), respectively.

Acemagic is offering a two-year warranty across all models, which is in line with consumer electronics standards but not unusually generous for a system designed to compete with higher-end business PCs.

Unlike many compact systems, the F5A includes two M.2 2280 PCIe 4.0 x4 slots, supporting up to 4TB of storage, along with memory expandability up to 128GB of DDR5.

The cooling system features copper heat pipes and dual fans, with added thermal management for solid-state drives.

While Acemagic emphasizes quiet operation, real-world performance under sustained load will be the true test of its effectiveness.

Connectivity is another area where the F5A appears well-prepared. It supports Wi-Fi 7, Bluetooth 5.4, and dual 2.5Gbps LAN ports.

For wired expansion, there are two USB4 ports delivering up to 40Gbps transfer speeds, along with HDMI 2.1, DisplayPort 2.1, and OCuLink.

The device also supports four simultaneous 4K displays, which could appeal to users seeking a compact workstation environment.

Pre-installed Windows 11 Pro comes as standard, with support for Linux distributions like Ubuntu and Manjaro for those who prefer open-source systems.

Acemagic’s F5A will compete with the recently released Geekom A9 Max, another HX370 model, which costs more.

While Geekom leans on a longer warranty and established brand recognition, Acemagic counters with affordability and flexible expansion.

The comparison reflects the wider dynamics of the mini PC market, where buyers weigh cost and specifications against support and brand assurance.

You might also like

• These are the best vertical monitors around today
• We've also listed the best laptops for working from home
• Windows 95 at 30 - Way ahead of its time, or the greatest Microsoft game-changer?