TechRadar News

• HPE pushes Morpheus as VMware virtualization costs rise under Broadcom
• Socket-based pricing offers cost predictability for high-core system deployments
• Morpheus aims to compete but VMware remains deeply rooted in enterprise

Hewlett Packard Enterprise has looked to address growing rumblings from its customers about rising VMware licensing fees following Broadcom’s acquisition of the virtualization company in 2023.

HPE is actively responding to these concerns, especially from those seeing major increases in virtualization costs, by pushing its Morpheus platform at its recent HPE Discover 2025 event.

HPE acquired Morpheus in 2024 and has since worked to develop it into a viable option for companies looking to reduce or end their reliance on VMware.

A choice of editions

Morpheus is being integrated into HPE’s broader Private Cloud offerings, with two editions available to suit different needs.

The VM Essentials edition is geared toward smaller environments and supports management of local KVM clusters and VMware clusters through vCenter.

This edition has a much lower entry cost, priced around $600 per socket, and unlike Broadcom’s per-core licensing, which some say penalizes scale, HPE’s per-socket model remains fixed regardless of hardware configuration.

This could appeal to those with high-core-count systems and help organizations better keep on top of their infrastructure costs.

For more advanced deployments, there’s Morpheus Enterprise. This includes profiling and cost calculator tools to help compare cloud and on-prem options. This version is priced at about $2,500 per socket.

HPE’s goal appears to be giving customers a clearer path away from VMware, starting small and scaling up.

Whether that’s enough to make a dent in VMware’s deep entrenchment across enterprise environments remains to be seen, but HPE is clearly making the case.

ServeTheHome's Patrick Kennedy writes, “We know many folks are struggling with virtualization licensing costs. It seems like HPE sees the need. It was actually neat to see that HPE is aggressively trying to move customers off of VMware and the Morpheus booth was certainly busy today. For folks who are still looking for solutions, this might be one to look into especially if you are a HPE shop.”

You might also like

• This is the best virtual machine software available right now
• Companies switching from VMWare should expect high-cost high-risk journey
• HPE now allows customers to run its most powerful server on AWS

Likeable brand Nothing has unveiled a set of striking over-ear headphones and if, like me, you're ready to fully embrace cans that aren't trying hard to look like the Sony WH-1000XM4 (and, by association, like every other set of headphones on the market), you'll probably be fully on board with these.

Might it appear as if I've got two cassette tapes lashed around my head – two, ahem, head cleaners, if you're old enough to get the gag? Quite possibly. Do I mind? Not one jot thank you very much; know me, know that I have a lot of love for the resurgence of the humble cassette format.

Anyway, I've worn them for a week now and sadly, owing to various time constraints, I wasn't able to compile TechRadar's full Nothing Headphone (1) review. However, I do share my colleague Matt Bolton's verdict and of course, that doesn't mean I have no further thoughts… the first being that the design here is a rare delight for the money.

Since the product has been under embargo until very recently, I've not been able to wear the Headphone (1) in public. But believe me, I cannot wait to do that. The transparent accents on the ear cups catch reflections and the light quite beautifully and there's a quirky, otherworldy-but-the-90s aesthetic to them. They remind me of Björk's 1997 All is Full of Love video – why wouldn't I want that?

(Image credit: Future)Nothing ventured, Nothing gained

So they look delightfully oddball, and you love to see it. They're also comfortable – for me, anyway, and I do wear glasses at work these days (see "old enough", above).

Furthermore, the Nothing X companion app is a joy. It's clean, it's stylish, it's easy to follow and it puts things like noise cancellation and head-tracked streaming-service agnostic spatial audio up front and center, exactly where you want those perks. Honestly, I love the app.

So where does it start to go wrong? My one minor gripe with the design is to do with the volume roller on the right ear cup. Don't get me wrong, I adore it; why has no other brand implemented anything even close, except for the Fairphone Fairbuds XL and Marshall Monitor III ANC, although both were different in that they were bijou joystick-style dials, rather than a roller?

No, my issue is the piped-in sound you hear when you use it. I know it's supposed to mimic a rotary watch bezel, but it's such a sweet, treble-centric whirr and so close to my delicate ear that it sets my teeth on edge (and there's no way to turn it off). In fact, even the power on buttons are a little sweet for my liking.

(Image credit: Future)Much ado about Nothing?

You're probably thinking: come on, this issue feels small-fry and you've already said you love the design, what's the big problem? And you'd be right: I think the bare bones here are truly exciting. Nothing goes its own way and I'm all for it. I've championed the Nothing Ear (a), the firm's triumphant third stab at earbuds, for over a year now.

The thing is, there's one key area where the Heaphone (1) don't quite pass muster at the level. Sonically, I pitted them against my trusty Cambridge Audio Melomania P100, and the circumaural soundstage of the P100 betters the Nothing option – something you hear straight away.

Nothing has gone to great lengths to tell us about the "custom 40 mm dynamic driver, built with high-linearity suspension and an 8.9 mm PU surround that moves air more freely than conventional PET materials" alongside a "nickel-plated diaphragm, including rim and dome".

Then, enter hi-fi giant KEF, which provided the rigorous tuning and testing of the headphones. And I think perhaps therein lies the rub. Nobody is challenging KEF's formidable audio chops, but maybe (just maybe) a modicum of function has had to be lost in the name of a finessed form?

KEF is a master of its trade, and clearly the intention here is serious, integrated, neutral and layered hi-fi sound. And this is very much on the way to being achieved – in fact, in acoustic mixes and with spatial audio deployed, I found admirable separation and insight. Sonically, they are not bad.

It's just that I've reviewed every set of Nothing earbuds to date, and I know the sound Carl Pei's company and its collaborative effort is capable of achieving, given a fresh iteration. The result here is a set of headphones that while relatively detailed, come off a little dynamically lean.

(Image credit: Future)If at first you don't succeed… 

Perhaps if the Headphone (1) release hadn't been so rigidly tied to the simultaneous unveiling of the Nothing Phone (3)? Perhaps if Nothing's design language hadn't been of such paramount importance, there may have been leeway to eke a little more out of the svelte drivers nestled in their much-adorned ear cups? I'm speculating – I am.

But I'm also urging Nothing and KEF to work more on this partnership, because this product as an inaugural effort is beautiful to me, and the sound is well on its way.

I listened to D'Angelo's Brown Sugar (I decided to give my Apple Music Replay All Time playlist a spin; a trip back to 2016 I'm not sure I needed) and while the vocals do dart out from shadowy corners in the mix, it's an all-together more immersive, dynamically agile listen when played through the cheaper Cambridge Audio rival.

Again, I want to urge Nothing to keep working on these. I truly admire the release of the Nothing Headphone (1) when most smartphone-centric firms are sticking to cheaper earbuds propositions alongside handsets. Case in point: Samsung, a heavy-hitter that hasn't launched a set of actual headphones in nearly 10 years, following mixed reviews for the 2014 Level Over and 2016 Level On.

In 2025, Nothing's got something here that could blow everything else away, and while the sound is not perfect yet, the design and feature-set is too good to shelve. Very few manufacturers ace a set of headphones at the very first time of asking, and to say that I eagerly await for the release of the Headphone (2) is an understatement.

(Image credit: Future)You may also like

• I'm an audio expert and these are the only 4 headphones and Bluetooth speakers I want to see in the Prime Day deals
• Everyone needs 5 pairs of headphones: these are the ones to buy
• See our pick of the best noise-cancelling headphones around

• Techland has discussed how the narrative is shaping up in Dying Light: The Beast
• The developer is doing away with branching choices
• This is in an effort to create a starting point for the series' future

Developer Techland will be stepping away from player-driven narrative choices for the upcoming Dying Light: The Beast. But it seems like there's going to be a very good reason for that, at least in relation to the game's returning protagonist - Kyle Crane.

In an interview with GamesRadar, Dying Light franchise director Tymon Smektała explained: "We wanted this to be a canon entry into the series. We're looking back at what happened to Kyle Crane in the first game, and we want to tell the next chapter of the story."

"One thing that Dying Light: The Beast does is hint at the future of the series," he adds. "We really wanted to make sure that at the end of the game, Kyle Crane is where we want him to be, so that's why we decided to back down from the narrative choices of Dying Light 2."

Furthermore, it's likely that Techland is looking to shape the future of the Dying Light franchise with what happens at the conclusion of The Beast.

While he didn't drop any juicy story hints, Smektała told TechRadar Gaming: "We definitely have, I would say, some Marvel [style] post-credit scenes that hint at some things that are going to happen in the future."

Smektała also provided further context on the decision to move away from branching narrative choices, adding: "If we would have given the players agency in terms of how they want to shape the story, then it will be harder for us to have Kyle where we want him to be at the end of the game.

"So this is a very canonical Kyle story that answers everything that happens to him before and after the events of Dying Light."

You might also like...

• Steam's new performance overlay feature will help you monitor how your games are running
• Monster Hunter Wilds Title Update 2 finally fixes the awful performance issues that have plagued PC players since day one
• If you're nostalgic for classic Guitar Hero and Rock Band games, you'll probably want to keep an eye on this modular guitar controller

• There will be a free Nintendo Switch 2 upgrade for Shadow Labyrinth
• The game will run at 4K 60fps on the Nintendo Switch 2
• It's available to pre-order now

Upcoming strange Pac-Man game Shadow Labyrinth will run at 4K 60 frames per second (fps) on Nintendo Switch 2 in docked mode thanks to a free Nintendo Switch 2 Edition Upgrade Pack.

This was revealed as part of the most recent trailer for the game, which gave us a decent look at the Nintendo Switch 2 version in action. It also presented some side-by-side footage showing the Nintendo Switch version too, which seems to be running at a noticeably lower frame rate and resolution.

Both the Nintendo Switch and Nintendo Switch 2 versions of the game are up for pre-order via the eShop, where they both cost $29.99 / £24.99. The Nintendo Switch 2 Edition Upgrade Pack is also available to pre-order and is completely free.

Shadow Labyrinth was first revealed at The Game Awards 2024, with a subsequent appearance in the Nintendo Switch 2 Direct. A darker take on Pac-Man, the game is a 2D metroidvania adventure in which the protagonist Swordsman No. 8 explores a sinister maze accompanied by the familiar yellow Puck.

It features plenty of combat and traversal, but also a few elements directly inspired by the classic arcade Pac-Man gameplay. It's set to launch on July 18 this year and is coming to PC, PlayStation 5, Xbox Series X and Series S in addition to Nintendo Switch and Nintendo Switch 2.

You might also like...

• I've replaced my Switch OLED with a Nintendo Switch 2 for two weeks now - here are the biggest improvements I’ve noticed in my time with the new console
• Donkey Kong Bananza offers a no-holds-barred open world that encourages you to go bananas
• I’ve clocked up more than 60 hours in Mario Kart World – here are my 3 top tips for getting first place

• You can now edit texts sent from Android to iOS
• The change is part of the updated RCS standard
• Texts sent from iOS to Android can’t yet be edited, though

Sending text messages between an iPhone and an Android device has long been a pretty poor experience, with features like typing indicators and read receipts missing for years. That’s shifted in recent years thanks to the use of Rich Communication Services (RCS), and it’s bringing another benefit to your cross-platform chats.

In this case, that’s the ability to edit texts sent from an Android phone to an iPhone (via Android Authority). This feature appears to be rolling out gradually to Android users, so it’s not available to everyone just yet. But if it’s working for you, all you’ve got to do is long press on a sent message, then tap the pencil icon, make your adjustments and save your message.

Unfortunately, this doesn’t work the other way around – that is, texts sent from an iPhone to an Android device cannot be edited. Presumably, Apple will need to update its Messages app to add support for this functionality.

You’ve been able to edit texts sent between iPhones for years, and messages going from one Android device to another have been editable when using RCS for about twelve months. But although editable messages are now part of RCS, companies like Apple and Google need to support the feature – which is why it’s not available in iOS right now.

Slowly adding support

(Image credit: Shutterstock / Tada Images )

Apple has been reluctant to support RCS for a long time, partly because it previously offered much weaker encryption than Apple’s iMessage platform, which is end-to-end encrypted. However, the change that introduced editable texts to RCS has now also brought forth end-to-end encryption, which might help to smooth things over with Apple.

The rollout of editable messages also hasn’t been entirely pain-free. While edited messages appear as normal on Android (with a small “Edited” timestamp underneath them), they behave differently in iOS. There, iPhone users see a second message preceded by an asterisk, doubling up the number of texts on their screen.

Both Apple and Google gave their support to cross-platform RCS messages earlier this year, so we’re hoping that these bugs and oddities will be ironed out in due course. For now, though, the situation when texting across phone platforms has been improved, even if only in one small way.

You might also like

• Apple will finally enable encrypted RCS messages between iOS and Android, and it's about time
• What is RCS messaging? The iMessage rival and SMS text replacement explained
• Hell freezes over – Apple to support RCS messages from Android phones next year

• The iPhone 17 Air will apparently be just 5.5mm thick
• That's even slimmer than the Samsung Galaxy S25 Edge
• However, it now looks very likely to only have one rear camera

The iPhone 17 Air is a rumored phone that – no matter what else might be good or bad about it – will in large part be judged on how slim it is. After all, being slim appears to be its entire USP. And on that front, it could be a success.

In a new YouTube video, FrontPageTech (via NotebookCheck) has claimed that the iPhone 17 Air will be just 5.5mm thick, which would make it even slimmer than the 5.8mm thick Samsung Galaxy S25 Edge. While that’s only a 0.3mm difference, this could still be a big win for Apple, since the S25 Edge would be this rumored phone’s main competitor, and is also sold on being slim.

Of course, we’d take this thickness claim with a pinch of salt, especially as FrontPageTech has a mixed track record. But lately they’ve been right more often than they’ve been wrong, and this leak is broadly in line with previous iPhone 17 Air leaks – though one puts it at a slightly slimmer 5.44mm and another at a marginally thicker 5.65mm. In all cases though, leaks suggest it will have Samsung’s rival phone beat on that front.

Compromised cameras and a Plus-level price

However, the iPhone 17 Air probably won’t have the Samsung Galaxy S25 Edge beat in every other way, as this latest leak – among many others – suggests the iPhone 17 Air will have just a single 48MP camera. That’s in contrast to a dual-lens camera on the S25 Edge, consisting of both a 200MP main sensor and a 12MP ultra-wide camera. So it seems Apple is probably prioritizing thinness over photography.

Other specs have also been leaked here, and again largely echo what we’ve heard before. This includes an A19 – but not A19 Pro – chipset, 12GB of RAM, and a 6.6-inch screen with a 120Hz refresh rate.

The video also highlights the battery, saying that Apple will use one with a silicon anode, in order to achieve a higher density than would otherwise be possible, and that this – coupled with an efficient modem and chipset – could mean the iPhone 17 Air will last almost as long as the base iPhone 17, despite the limited space for a battery.

Finally, the video touches on price, stating that the iPhone 17 Air will cost roughly the same amount as the iPhone 16 Plus – a phone that starts at $899 / £899 / AU$1,599. This too is in line with multiple previous leaks.

We should find out whether this is all correct or not in September, as that's when the iPhone 17 Air is expected to launch – alongside the rest of the iPhone 17 series.

You might also like

• The foldable iPhone’s cameras might leave it trailing behind the Galaxy Z Fold 7
• The iPhone 17 Pro’s rumored camera bar looks much better than before in new dummy unit photos
• iPhone 17 Pro again tipped to get a vapor chamber – here's why that'd be a big deal

Even with today’s vast arsenal of cybersecurity tools and AI-enhanced threat detection, attackers continue to succeed – not because the technology is failing, but because the human link in the defensive chain remains exposed. Cybercriminals almost always take the path of least resistance to execute a breach, which often means targeting people rather than a system.

According to McKinsey, a staggering 91% of cyberattacks have less to do with technology, and more to do with manipulating and taking advantage of human behavior. In other words, despite technologies like AI advancing at break-neck speed, cybercriminals are still more likely to hack people than machines.

From a cybercriminal’s perspective, this makes sense. It’s the path of least resistance. Why spend resources hacking your way through a high-tech, AI-secured front door when there’s an open window around the back? This isn’t news to CISOs – according to a 2024 IBM survey, almost three-quarters (74%) now identify human vulnerability as their top security risk. They’re aware of the open window, and now they’re trying to secure it.

Easier said than done

That’s easier said than done, however. Whether it’s a well-timed phishing email, a spoofed call, a deepfake video, or a barrage of authentic-seeming push notifications designed to wear down a user’s judgment, attackers are adapting faster than defenses can compensate.

The reality is that while security vendors race to outpace attackers with smarter algorithms and tighter controls, the tactics that most reliably lead to breaches are psychological, not technical. Threat actors are exploiting trust, fatigue, social norms, and behavioral shortcuts – tactics far more subtle and effective than brute-force code.

It’s not a lack of technology leaving organizations vulnerable to these techniques, it’s a lack of alignment between those tools and the way people actually think and operate. In fast-paced, high-pressure environments, employees don’t have the bandwidth to second-guess every request or scrutinize every prompt.

They rely on instincts, familiarity, and patterns they’ve learned to trust. But those very instincts are what attackers hijack, turning help desk tickets into access exploits, or mimicked CFOs into multi-million-dollar heists. As generative AI accelerates the realism and reach of these tactics, organizations face a critical question: not just how to keep the bad actors out, but how to better equip their people within. Because when breaches hinge on human decisions, cybersecurity isn’t just a technology issue – it’s a human one.

Trust, bias, and the psychology of security breaches

Human behavior is a vulnerability, but it’s also a predictable pattern. Our brains are wired for efficiency, not scrutiny, which makes us remarkably easy to manipulate under the right conditions. Attackers know this and design their exploits accordingly. They play on urgency to override caution, impersonate authority figures to disarm skepticism, and drip-feed small requests to trigger consistency bias. These tactics are ruthlessly calculated, and they work not because people are careless, but because they’re human.

In early 2024, a finance worker at a Hong Kong firm was tricked into transferring $25 million after attending a video call with what appeared to be the company’s CFO and other colleagues – each one a convincing AI-generated deepfake. The attackers used publicly available footage to clone faces and voices, creating a seamless illusion that exploited trust and familiarity with devastating effect.

The eye-opening part is that these deepfake tools are now readily available. Modern social engineering doesn’t rely on obvious red flags. The emails aren’t riddled with typos, and the impersonations don’t sound robotic. Thanks to generative AI, deepfake technology, and access to vast training data, attackers can now create incredibly convincing personas that mirror the tone, behavior, and language of trusted colleagues. In this environment, even the most well-trained employee can fall victim without fault.

Heuristics – mental shortcuts – are frequently exploited by attackers who know what to look for. “Authority bias” leads people to follow instructions from perceived leaders, like a spoofed email from a CEO. The “scarcity principle” ramps up pressure by creating false urgency, making employees feel they must act immediately.

And “reciprocity bias” plays on basic social instincts – once someone has received a seemingly benign gesture, they’re more likely to respond positively to a follow-up request, even if it’s malicious. What so often looks like a lapse in judgment is often just an expected outcome of cognitive overload and the common, everyday use of heuristics.

Where policy meets psychology

Traditional identity and access management (IAM) strategies tend to assume that users will behave predictably and rationally – that they’ll scrutinize every prompt, question every anomaly, and follow policy to the letter. But the reality inside most organizations is far messier. People work quickly, switch contexts constantly, and are bombarded with notifications, tasks, and requests.

If security controls feel too rigid or burdensome, users will find workarounds. If prompts are too frequent, they’ll be ignored. This is how good policy gets undermined – not out of negligence, but because the design of the system clashes with the psychology of its users. Good security mechanisms shouldn’t add friction; they should seamlessly guide users towards better choices.

Applying principles like Zero Trust, least privilege, and just-in-time access can dramatically reduce exposure, but only if they’re implemented in ways that account for cognitive load and context. Automation can help here: granting and revoking access based on dynamic risk signals, time of day, or role changes without requiring users to constantly make judgment calls.

Done right, identity management becomes an invisible safety net, quietly adapting in the background, rather than demanding constant interaction. Humans shouldn’t be removed from the loop, but they should be freed from the burden to catching what the system should already detect.

Building a security culture

Technology may enforce access policies, but culture determines whether people follow them. Building a secure organization has to be about more than simply enforcing compliance. That starts with security training that goes beyond phishing drills and password hygiene to address how people actually think and react under pressure. Employees need to recognize their own cognitive biases, understand how they’re being targeted, and feel empowered – not penalized – for slowing down and asking questions.

Equally important is removing unnecessary friction. When access controls are intuitive, context-aware, and minimally disruptive, users are more likely to engage with them properly. Role-based and attribute-based access models, combined with just-in-time permissions, help reduce overprovisioning without creating frustrating bottlenecks in the form of pop-ups and interruptions. In other words, modern IAM systems need to support and empower employees rather than make them constantly jump through hoops to get from one app or window to another.

The human firewall isn’t going anywhere

The biggest takeaway here is that cybersecurity isn’t just a test of systems, AI-driven or not – it’s a test of people. The human firewall is arguably an organization’s biggest weakness, but with the right tools and policies in place, it can become its greatest strength. Our goal should not be to eliminate human error or change the innate nature of humans, but to design identity systems that make secure behavior the default – easy, intuitive, and frictionless.

We list the best employee recognition 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

Full spoilers immediately follow for Ironheart.

Ironheart has disengaged its thrusters. Its final three episodes landed on July 1 or 2 (depending on where you live), with the trio's arrival marking the end of the Marvel TV show's run on Disney+.

Or does it? The last Marvel Phase 5 project's finale is full of unresolved plot threads that suggest Riri Williams (Dominique Thorne) and Parker Robbins' (Anthony Ramos) stories have only just begun in the Marvel Cinematic Universe (MCU). Throw in the debut of a long-awaited Marvel villain and other interesting tidbits that might impact the MCU moving forward, and Ironheart could be more important to the MCU than we realized.

But enough babbling. Let's take a look at the biggest questions you have about Ironheart's ending and I'll do my best to answer them.

Who plays Mephisto in Ironheart? And what is he, exactly?

The demon lord known as Mephisto in Marvel Comics has finally made his MCU debut (Image credit: Marvel Studios/Disney+)

Let's address the second question first because it's easier to answer: Mephisto is played by Sacha Baron Cohen – he of Borat, Ali G, and The Trial of the Chicago 7 fame – in the MCU.

Cohen's appearance won't shock some MCU devotees. He was first linked with the role of Mephisto in October 2022 – those reports arriving 18 months after Marvel fans had convinced themselves Mephisto would be unveiled as primary antagonist of WandaVision. As it happened, Agatha Harkness was that show's Big Bad. Nonetheless, fans have longed for the day that Mephisto would finally make his bow in Marvel's cinematic juggernaut.

As for who (or what) Mephisto is: he's Marvel Comics' takes on Mephistopheles, the demon lord who appears in the Germanic fable Faust. Mephisto is also an alias of The Devil, aka the mythical personification of evil who lauds it over the realm known as Hell in many religions.

Mephisto's true form can be seen in a easily-missed shot when he taps a teaspoon on the side of a cup of coffee (Image credit: Marvel Studios/Disney+)

But back to Marvel's take on Mephisto. First introduced in comic book 'Silver Surfer #3' in December 1968, the demonic entity is a recurring adversary of The Silver Surfer, Ghost Rider, and Spider-Man in Marvel literature, although he's also crossed paths with Doctor Strange, Doctor Doom, Scarlet Witch, and myriad other Marvel heroes and villains.

He's most famous for being responsible for the demise of Peter Parker and Mary Jane Watson's marriage, too. Indeed, in the incredibly divisive 2000s comic story 'Spider-Man: One More Day', Mephisto made a deal with the pair to save Aunt May's life in exchange for the formal union of their love, which infuriated long-time fans of the webslinger.

Abilities wise, Mephisto is unsurprisingly one of the most powerful beings in the Marvel Universe. He possesses all of the usual superhuman powers, but it's his ability to shapeshift, foresee events before they transpire, and prey on people's biggest fears and/or their trauma that makes him such a threat.

It's the latter of those skills, coupled with his vast magical powers, that enables him to warp reality, and to manipulate people and bargain with them for something they want. Usually, the trade-off is the other individual's soul, which Mephisto takes ownership of once a deal is formally agreed.

Okay, so what's the deal with him first appearing in Ironheart? He's the individual who rescues Parker Robbins after the latter and Cousin John's unsuccessful break-in at the home of Parker's dad. Then, he convinces Parker to strike a deal: Mephisto will gift him a dark magic-infused hooded cloak – so Parker can commit petty crimes and make himself rich – in exchange for Parker's soul.

Who dies in Ironheart?

Ironheart performs the rare feat of bringing someone back from the dead (Image credit: Marvel Studios/Disney+)

Nobody. The eponymous hero survives, as do Anthony Ramos' villain in Parker Robbins/The Hood and Alden Ehrenreich's secondary antagonist Zeke Stane.

In fact, Ironheart pulls a reverse Uno card, and brings someone back from the dead. After she defeats The Hood by taking away the source of his powers (aka his cloak), Riri heads for the exit of Desperito's, aka the dilapidated pizza restaurant that The Hood turned into his hideout.

On the way, though, she bumps into Mephisto, who cajoles her into becoming his next victim. Mephisto makes a deal with Riri to revive Natalie Washington (Lyric Ross). Remember, Natalie is Riri's best friend, who was killed in a drive-by shooting years ago, and who N.A.T.A.L.I.E, the artificial intelligence Riri creates, is based on.

In return, Mephisto gets Riri's soul. Indeed, after Riri shakes hands with Mephisto, we're privy to a highly emotional reunion between Riri and the reborn Natalie. However, as the pair hug, Riri's right forearm gets covered by the same fire-esque skin markings that adorned Parker's body every time he used his cloak and/or powers.

With Riri's soul now seemingly bound to Mephisto, it'll be fascinating to see when she realizes this is the case, and if her and/or Natalie sacrifice the latter's second chance at life to free Riri of the pact that this flawed hero has made.

Does Ironheart have a mid-credits or post-credits scene?

Parker and Zelma meet in Stanton's in Ironheart's mid-credits scene (Image credit: Marvel Studios/Disney+)

Yes and no. Ironheart has a mid-credits stinger, but there's no end-credits scene to stick around for.

Where the former is concerned, we see the now-cloakless Parker visit Stanton's. That's the confectionary shop that's a front for the Stanton family, who deal with the mystical and the occult.

Anyway, upon entering the store, Parker is greeted by Zelma Stanton (read my Ironheart cast and character guide for more on her), who rattles off the welcome spiel we first heard in episode 4. Parker, though, immediately sees through Zelma's facade, and informs her that he knows about the stuff in the back – i.e., the magical trinkets and knowledge in the Stantons' possession. Parker then tells Zelma he's looking for someone more "experienced" and "supreme" than she is before the screen cuts to black.

Clearly, Parker is interested in finding one of two individuals: Stephen Strange, aka Doctor Strange, or the current Sorcerer Supreme/Master of the Mystic Arts in MCU fan-favorite Wong. Considering that Strange is off-world dealing with the fallout from Doctor Strange 2, I suspect Parker wants to find Wong to learn more about magic. Will it be for the right reasons because he's turned over a new leaf in his life, or does he seek more knowledge of dark magic in order to exact revenge on Riri and Mephisto? Who knows – but this isn't the last we've seen of Parker.

Will there be an Ironheart season 2? And how could it set up Marvel's Strange Academy TV show?

I'd be surprised if Ironheart gets a second season (Image credit: Marvel Studios/Disney+)

We don't know. Unlike their big-screen offerings – read my articles on how to watch the Marvel movies in order and the best Marvel movies while you're here – Marvel rarely tells us if a show's lead character will return in a second season once its final end credits sequence has rolled. The only time I can remember the comic giant doing so was with season 1 of Loki, with a mid-credits stinger stating "Loki will return in season 2".

The Tom Hiddleston-starring series aside, Marvel has only renewed three other shows: Daredevil: Born Again, whose second season is currently in development, and animated offerings Your Friendly Neighborhood Spider-Man and What If...?. The former's sophomore season is also in active development, while the latter ran for three seasons between 2022 and 2024.

The chances of Ironheart season 2 being greenlit, then, aren't too high. Sure, its 86% critical score on Rotten Tomatoes suggest it should be, but its mixed audience rating indicates that general viewers wouldn't be devastated if a second season isn't forthcoming.

In my view, it's far more likely that Williams, Robbins, and/or Mephisto will show up in other MCU projects. Williams could be part of the cast for Marvel's long-rumored Young Avengers project. Based on Ironheart's mid-credits scene, Robbins could show up in another rumored Disney+ show called Strange Academy, which could star Benedict Wong's Sorcerer Supreme and explore the Mystic Arts in more detail. As for Mephisto, given his shared history with Spider-Man, would it be too much to ask for a cameo in Spider-Man: Brand New Day? Probably, but stranger things have happened.

You might also like

• I watched Ironheart on Disney+, and Marvel and Ryan Coogler have conjured up a magical show that falls just short of being iconic
• 'It changed our release plan': Ironheart producers explain why it's taken so long to come out on Disney+
• These are the best Disney+ shows to stream today

Modern IT systems are awash with a constant flow of data providing information on system performance and security. This data is crucial for IT professionals looking to keep IT infrastructures running smoothly.

The snag is, with so much information available, it’s almost impossible to decipher what’s routine and what potentially could pose a threat. Trying to figure out the digital wheat from the chaff requires more than just visibility. It requires increasingly sophisticated systems that can interpret, prioritize and act—not simply collect information.

Unfortunately, most observability tools today don’t work this way. They generate alerts, log events and surface anomalies. And while technology is improving all the time, they don’t always understand what’s happening or how to respond.

Take a typical global enterprise running a hybrid architecture. It may have critical applications that are spread across multiple cloud providers, while also relying on on-premise legacy systems. Together, these systems are shepherded by dozens of monitoring tools generating thousands of alerts every day.

Some are false positives. Others are minor policy breaches. But lurking somewhere in the noise is a genuine security incident. And by the time it’s spotted… well, it could be too late.

Observability is becoming more intelligent

What’s missing isn’t just more tools or greater visibility. It requires a new, highly sophisticated level of observability—one that works more like a human brain with the ability to filter out noise, recognize what matters and trigger the right response at the right time. What’s needed is something intelligent that can ’think’ for itself.

Part of the reason why this is needed is because IT teams have tended to invest in separate tools that often have little contextual awareness. That means it’s up to the human members of IT teams to bridge the gaps, deciding whether an alert is serious, identifying the root cause and initiating the right response. In fast-moving environments, these human assessments can take time which, in turn, adds to the risk.

An intelligent observability system, on the other hand, would do more than simply monitor for known issues. It would detect anomalies in real time using context-aware monitoring, then assess the severity and likely impact based on both the technical and business relevance and risk.

Rather than treating every signal the same, it would prioritize based on urgency and risk, helping teams focus on what truly matters.

Crucially, it would also support automation, enabling routine fixes or containment measures to be initiated. And instead of splitting insight across multiple disconnected views, it would bring together data from on-premises and cloud environments in a single, cohesive picture.

This kind of system doesn’t just monitor IT systems and networks. It has total oversight and is ready to act when needed.

So, how close are we?

The good news is that progress is being made. AI-driven observability is moving from aspiration to implementation. Anomaly detection based on behavioral baselines is becoming more accessible, helping teams distinguish real issues from false alarms. Alert correlation and intelligent escalation paths are improving, reducing alert fatigue and bringing the right signals to the right people at the right time.

Some observability platforms, including those being developed at SolarWinds, are already combining monitoring, analysis and response into more cohesive workflows. Integration across hybrid environments remains a challenge, but the building blocks for intelligent observability are now in place.

What’s still missing, though, is the kind of full-system intelligence that can replicate the nuance of human decision-making. Most observability tools still rely on thresholds, templates, or predefined rules. True context-awareness—the ability to understand why something is happening and what to do next—is still emerging. But the direction of travel is clear.

Why this matters now

According to a recent SolarWinds AI and Observability report focused on the public sector, three-quarters of respondents said hybrid environments were hard to manage. Top concerns included data protection, integration complexity and a lack of visibility across systems.

Managing this complexity is made harder by the reality that observability tools are often siloed—one for cloud, another for on-prem, with separate platforms for detection, logging and remediation.

Security only adds to the unpredictability. In the same report, more than half of IT professionals said insider mistakes were contributing to serious threats, while 59% highlighted increasingly sophisticated attacks from external actors. The rise of generative AI means those external threats are becoming more scalable and targeted, increasing the strain on overstretched IT teams.

Which is why the key is not to add even more tools but to reduce complexity, improve visibility and act with intelligence and speed. An observability system that functions more like a brain does exactly that, because IT systems need to do more than observe. They need to understand.

We list the best small and medium business (SMB) firewall 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

• Report finds business AI adoption is exploding, but most companies are skipping the hard work of preparation
• Leadership teams are failing to align on AI priorities, leaving strategies fractured and confused
• AI is only as good as the data behind it, and most data strategies are missing

The surge in artificial intelligence adoption has sparked comparisons to the cloud boom of the past decade, but while usage is growing rapidly, understanding remains shallow, new research has claimed.

A Hostinger report found almost 80% of companies now use or plan to use AI, but a seperate Adecco Group report claims only 10% of C-suite leaders believe their organizations are fully ready for the disruption AI brings.

Among the estimated 359 million companies worldwide, about 280 million now integrate AI into at least one function.

AI adoption accelerates, but strategy and structure lag behind

A growing number of small businesses are turning to the best AI tools to handle tasks like writing emails, analyzing data, or generating content.

Larger companies may build out full teams for implementation, but smaller firms are quietly transforming operations using lean, sometimes improvised, approaches.

Still, readiness doesn’t follow adoption, and there is a worrying gap in strategy, as although 60% of leaders expect workers to update their skills, 34% of companies have no formal AI policy.

Adecco found over half of CEOs admit their teams struggle to align on priorities, and only a third of businesses are investing in data infrastructure that would help close these gaps.

However, a small group of “future-ready” companies is building more responsive strategies by supporting continuous learning and relying on enterprise-wide insight to shape their AI direction.

Adecco’s CEO, Denis Machuel, puts it plainly: “AI-driven transformation must be human-centric.”

Many companies rush into AI adoption without understanding what differentiates them, resulting in scattered or redundant projects.

“Without enterprise-wide insight, AI efforts become siloed and misaligned. Enterprise Architecture can help focus AI initiatives on what truly sets a company apart,” Stendera explains.

By mapping their unique strengths and workflows, organizations can guide AI deployments that reinforce strategic priorities rather than dilute them.

AI depends not just on investment, but on introspection, and it is not a magic fix - and if companies do not understand what they need from AI, they won’t know how to use it, and the result will be catastrophic.

You might also like

• Check out the best AI phones on the market
• Here is our list of the best AI website builders on the web
• This AI tool claims to boost student grades by 15% using a 2400-year-old technique

• Streams 4K video, backs up your phone, and still skips cloud storage entirely
• The TerraMaster F4 SSD promises home privacy, but may offload all maintenance responsibility onto the user
• Up to 32TB of SSD speed sounds impressive, but your router might bottleneck it

As local storage continues to evolve, some brands now offer compact network-attached systems that emphasize privacy, speed, and media versatility.

The TerraMaster F4 SSD provides support for up to 32TB of SSD storage using four 8TB SSDs, and supports file systems such as EXT4, BTRFS, exFAT, and NTFS.

It bypasses the need for cloud-based platforms by incorporating hardware-level encryption and data segregation for over 20 user accounts.

Performance metrics meet household use cases

The device is powered by a quad-core ARM-based Rockchip RK3568 processor clocked at up to 2.0GHz, with hardware decoding support for H.264 and H.265 codecs and resolutions up to 4K@60fps.

It also features 8GB of DDR4 RAM, expandable to 32GB using two SODIMM slots.

Network capabilities include a 2.5GbE port and a 10Gbps USB 3.2 Gen 2 Type-C interface, while HDMI 2.0 offers 4K display output.

TerraMaster F4 SSD can also stream to TVs and tablets using standard protocols like uPnP and DLNA, and support for media servers such as Plex, Jellyfin, and Emby suggests it will be compatible with the majority of home setups.

The F4 SSD is built around a 5G Ethernet port, reportedly offering speeds up to five times faster than standard Gigabit connections, supporting high-throughput scenarios.

However, in practical terms, users will likely be limited by the speed of their broader home network, not just the NAS hardware.

Software features include backup tools (including cloud sync and snapshot), AI photo management, VPN server, and remote access via TNAS.online, which enables downloads and uploads through the cloud.

Security features include TRAID, a flexible array system that optimizes space while providing redundancy.

It also supports RAID 0/1/5/6/10, JBOD, and includes tools such as S.M.A.R.T., bad block scan, SSD trim, and hot spare management.

TerraMaster’s SPC control system is another layer that restricts app access based on verified permissions.

The F4 SSD also allows bi-directional syncing with cloud platforms like Google Drive and Dropbox.

This hybrid capability, while useful, may seem counterintuitive in a product designed to replace cloud reliance.

The NAS also features tool-free SSD installation with a drawer-style enclosure, making drive upgrades accessible to beginners.

Cooling is handled by a quiet convection fan system, reportedly keeping standby noise around 19dB, like the TerraMaster D4 SSD. Such silence may be beneficial in noise-sensitive environments like home studios or bedrooms.

The system operates on TOS 6.0 and supports up to 128 user accounts, 128 user groups, and 8 shared folder sync tasks.

This makes it suitable for advanced home users or small studios needing high-speed, private data access.

The TerraMaster F4 SSD is ambitious in scope, but its value will depend on whether users make full use of its features.

Via TechPowerUp

You might also like

• These are the fastest SSDs you buy right now
• Take a look at some of the best external hard drives
• Acer is giving away $240 worth of Google goodies for free with its Chromebooks

• Experts warn emails sent with sensitive data are still getting delivered unencrypted, and no one gets notified
• Microsoft 365 sends email in plain text when encryption fails, without alerting the user at all
• Google Workspace still uses insecure TLS 1.0 and 1.1 without warning senders or rejecting messages

Most users assume that emails sent through cloud services are encrypted and secure by default, but this might not always be the case, new research has claimed.

A report from Paubox found Microsoft 365 and Google Workspace both mishandle these failures in ways that leave messages exposed, without notifying the sender or logging the failure.

“Using obsolete encryption provides a false sense of security because it seems as though sensitive data is protected, even though it really is not,” Paubox said.

Default settings quietly undermine encryption

The problem isn’t just a technical edge case; it stems from how these platforms are designed to operate under common conditions.

Google Workspace, the report found, will fall back to delivering messages using TLS 1.0 or 1.1 if the receiving server only supports those outdated protocols.

Microsoft 365 refuses to use deprecated TLS, but instead of bouncing the email or alerting the sender, it sends the message in plain text.

In both cases, the email is delivered, and no warning is issued.

These behaviors pose serious compliance risks, as in 2024, Microsoft 365 accounted for 43% of healthcare-related email breaches.

Meanwhile, 31.1% of breached healthcare entities had TLS misconfigurations, despite many of these organizations using “force TLS” settings to meet compliance requirements.

But as Paubox notes, forcing TLS does not guarantee encryption using secure versions like TLS 1.2 or 1.3, and fails silently when those conditions are not met.

The consequences of silent encryption failures are far-reaching - healthcare providers routinely send Protected Health Information (PHI) over email, assuming tools like Microsoft 365 and Google Workspace offer strong protections.

In reality, neither platform enforces modern encryption when failures occur, and both risk violating HIPAA safeguards without detection.

Federal guidelines, including those from the NSA in the US, have long warned against TLS 1.0 and 1.1 due to vulnerabilities and downgrade risks.

Yet Google still allows delivery over those protocols, while Microsoft sends unencrypted emails without flagging the issue.

Both paths lead to invisible compliance failures - in one documented breach, Solara Medical Supplies paid more than $12 million after unencrypted emails exposed over 114,000 patient records.

Cases like this show why even the best FWAAS or ZTNA solution must work in concert with visible, enforceable encryption policies across all communication channels.

“Confidence without clarity is what gets organizations breached,” Paubox concluded.

You might also like

• These are the best VPNs with antivirus that you can use right now
• Take a look at our pick of the best internet security suites available
• China-backed "LapDogs" hackers hijacked hundreds of devices

If you're getting impatient waiting for Apple to release its long-promised AI upgrade for Siri, well, Apple might agree. There have been reports that Apple is talking with OpenAI and Anthropic about letting one of them power the next generation of Siri.

This would be truly surprising considering how close Apple usually keeps its tech development. If AI plays as big a role as Apple has implied it will in the future, this could be the foundation for a very different company.

And that is something I hope happens. It's not about cheering for the downfall of Apple’s in-house pride. I want this to happen not because I think Apple is incapable of building something great, but because I’ve been living with Siri for over a decade, and at this point, it feels like an ancient Dictaphone compared to communicating with today's AI chatbots.

Even with all the Apple Intelligence announcements this year, Siri still feels more like a feature than a full personality. You can ask her to set timers, maybe control your lights, but be careful how you speak or how conversational you get if you want it to still carry out your requests. By comparison, ChatGPT and Claude seem far quicker, more adaptable, and better at both understanding what people say and communicating back.

If Apple wants to finally make Siri feel like an assistant and not just a vaguely robotic concierge, borrowing someone else’s brain is a smart move. Apple and its Foundation Models group have been working on large language models for a while now, but progress has been mostly invisible to the average consumer. Breakthrough AI features, like contextual awareness and multi-step reasoning, just aren’t there yet.

While Apple is famous for taking its time to get it right (except for the Apple Maps fiasco), the rest of the world isn’t waiting. Google is stuffing Gemini into everything, while Microsoft has ChatGPT running in Office. Even Meta has Meta AI chat available on Instagram. Siri is still not where you go if you have follow-up questions about anything.

Siri AI

Sometimes, the smart thing is to stop fiddling with your own blueprint and just lease the engine. And these engine options are both pretty powerful. OpenAI’s GPT-4o can carry on fluent conversations, understand tone, and juggle complex context. Anthropic’s Claude is similarly capable, albeit with a reputation for more caution and focus on safety concerns, something Apple reportedly likes. Both companies are at the top of the field. Either one would instantly give Siri a massive IQ boost.

Of course, there are trade-offs. Apple loves control. Letting someone else power Siri means giving up a little bit of that control. What happens if OpenAI changes its pricing model? What if Anthropic gets cozy with Amazon, their big investor? What if there’s a data breach?

These are valid concerns. But Apple’s already shown it can integrate external AI in a way that keeps user data private. Their recent Apple Intelligence rollout uses something called Private Cloud Compute, which lets your data get processed on Apple’s own secure servers. Even when they do use ChatGPT for certain features, you have to opt in, and your requests aren’t stored.

Rumor has it Anthropic wants several billion dollars a year to license its tech. That’s not pocket change, even for Apple.

Rumor has it Anthropic wants several billion dollars a year to license its tech. That’s not pocket change, even for Apple. But it’s a lot cheaper than spending the next five years playing catch-up, especially if your catch-up product still doesn’t do what people expect from AI in 2025.

Apple could still develop its own AI models behind the scenes. It could roll them out when they are ready, while giving Siri a much-needed upgrade now. But, for most people, it doesn't matter who built the AI brain, just that it works. If Siri can maintain a conversation and respond to different ways of phrasing something, that would be a big step up. If most people ask Siri to make a reservation at a restaurant and it happens without any stumbles, they're not going to say, “Wait, was this answer generated by Anthropic or OpenAI?” They’ll simply accept that it worked and move on.

So yes, let Siri team up with a working AI model. All these iPhones could use the IQ boost. I don’t need Siri to be uniquely Apple or unique at all. I just need it to be useful.

You might also like

• Apple is about to make Personal Voice faster and better, and update almost all of its other accessibility features
• The ultimate AI search face-off - I pitted Claude's new search tool against ChatGPT Search, Perplexity, and Gemini, the results might surprise you
• 4 ChatGPT features I hope Apple adds to Siri
• Apple Intelligence feels like the HomePod all over again

• Google Gemini can more easily access your Phone and Messages apps
• Previously, you had to allow Gemini Apps Activity, reducing your privacy
• That's no longer the case, and comes as Gemini RCS support leaks too

Google looks set to give Google Gemini some serious upgrades by allowing it greater access to your apps without forcing you to tweak your privacy settings, and giving it new abilities within those apps.

You might have seen an email that Google sent to Android phone users stating that Gemini will be able to interact with the Phone, Messages, WhatsApp, and Utilities apps, even if you have Gemini Apps Activity switched off.

The immediate reaction was one of concern, but this is actually a huge privacy win.

(Image credit: Philip Berne / Future)

With Gemini Apps Activity switched on, Google can see an activity log of how you use Gemini – personal data it can then use to develop its products further. If you want to keep this data more private, you had to lose out on Gemini being able to access extensions, which allow it to perform actions using other apps, such as texting someone if you ask it to.

This change means you can keep your activity log private while still not losing out on these basic smart assistant features, which Google’s Assistant has had for years.

Though that’s not to say Google won’t store any of your Gemini activity even if this setting is off. Google admits it will store some activity data for at most 72 hours. It’s stored for 24 hours within Gemini so that the AI can respond to your conversations contextually. The longer limit is for security and safety reasons, which you can find out more about on Google’s support page.

RCS support incoming?

Perhaps in preparation for Gemini having easy access to Messages – and so presumably more people using the app – Gemini is getting an RCS upgrade too, apparently.

(Image credit: Apple)

That’s based on Android Authority’s analysis of the latest Google app files, which hints at RCS coming to Gemini because the AI can fetch the device’s RCS capabilities. This would only be necessary if the AI were RCS compatible.

This is significant because currently, Gemini’s inability to use RCS means it’s unable to send or play audio, images, or video through the Google Messages app. With access to this messaging standard, that could very quickly change.

As with similar leaks, there’s no guarantee that Gemini will get RCS support any time soon (or at all), but it certainly makes plenty of sense as an upgrade, so it’s one we’ll be keeping our eye out for.

You might also like

• This simple trick gets you 3 months of Google Gemini Veo 3 for free
• I tried Google’s new Gemini-powered clothing app – here’s how you can use AI to find the perfect outfit
• I've been using Google's new AI mode for Search – here's how to master it

• Anker just issued two separate recalls for some of its best power banks, spanning a total of six models
• Some of the power banks might pose a fire hazard
• If you qualify, you might be able to get a brand new power bank for free

Anker has just announced its second product recall this year – and while it's bad news for the brand, there could be a silver lining for you. If you own one of Anker's power banks, you might be eligible for a free replacement.

The company is recalling six different power bank models, citing issues with the lithium-ion battery inside. This includes models from Anker's Power Core, MagGo, and Zolo series, among others.

Below, we'll tell you everything you need to know about Anker's power bank recalls and what to do next.

Which Anker power banks are affected by the recall?

Anker has launched two separate recalls. One in early June included a single power bank model, while the more recent second expanded the list by an additional five models across a couple of Anker's product lines.

The power bank that was recalled first was the Anker PowerCore 10000 (A1263). If you bought the power bank through Amazon, you'd have received an email like the one we received below. The subject line would have been 'Attention: Important Safety Notice about your Past Amazon Order'.

(Image credit: Anker)

Not every PowerCore 10000 power bank was recalled. The recalled units only include the A1263 model, manufactured between January 1, 2016, and October 30, 2019. All of the recalled units were sold in the United States between June 1, 2016, and December 31, 2022.

The second recall was more comprehensive and included the following models:

• Anker Zolo Model A1681 (20K, 30-watt, with built-in USB-C and Lightning cables)
• Anker Zolo Model A1689 (20K, 30-watt, with built-in USB-C cable)
• Anker MagGo Model A1652 (7.5-watt, 10,000mAh)
• Anker Power Bank Model A1257 (10K, 22.5-watt)
• Anker Power Bank Model A1647 (22.5-watt, 20,000mAh, with built-in USB-C cable)

In these cases, Anker hasn't specified a manufacturing or sale date for the affected power banks. But in its 'voluntary recall' notice referring the above models, it stated that "while the likelihood of malfunction is considered minimal, out of an abundance of caution, we have decided to initiate a voluntary global recall of several Anker power bank models."

How to check if your Anker power bank has been recalled

(Image credit: Anker)

No matter which of the affected models you own, your first step will be to check the model and the serial number on the back or side of your power bank.

Not every unit from each model is being recalled, so even if you own one of these, it may not be eligible for replacement.

Next, you will have to input the serial number, along with a few other key details, on Anker's online recall form in order to check eligibility.

If it's confirmed to be one of the models affected, you should stop using it immediately and submit a product recall form. This contains information on how to request a gift card or replacement.

What can you get if your power bank qualifies?

If your power bank is affected by one of the recalls, Anker will either offer you a replacement or a gift card. Folks in the U.S. won't be receiving refunds, although Anker is offering refunds in China.

(Image credit: Anker)

In the case of the Power Core 10000, Anker says that you'll be offered a replacement power bank. If it's on back order, you can either choose a $30 gift card instead (to be used at Anker's store) or wait for the Power Core 10000 to be available again.

The second recall is slightly more vague, but Anker says that you'll be able to choose between a gift card or a replacement unit.

How to submit a recall claim

Once you're sure that you own one of the models listed above, submit your recall claim to Anker for verification.

• If you have the Anker Power Core 1000 (A1263), head to Anker's recall form for that model
• If you have one of the other power banks (A1257, A1647, A1652, A1681, A1689), go to this recall form instead.

In both cases, you may be asked for the serial number, proof of purchase, order number, and photos of the power bank.

If this fails, but you still think you're eligible for replacement, contact Anker's Support over live chat, email or phone.

Why are Anker power banks being recalled?

Anker hasn't gone into a whole lot of detail as to the cause of the recall, but in both cases, it seems that the problem lies with the battery.

When announcing the first recall, Anker said that these power banks could be a fire hazard due to a "potential issue with the lithium-ion battery." Reportedly, the battery can overheat, and if that happens, some of the plastic components might melt, resulting in smoke or even a fire.

In the case of the second recall, Anker revealed that it identified a potential issue with lithium-ion battery cells made by one of its vendors. The company admits that problems are unlikely, but it's being cautious and recalling all five of the affected models.

Can you still use your old power bank?

Once it's confirmed that your power bank is affected by the recall, you should stop using it immediately. Even if it seems to be working just fine, it poses a fire risk, so you should dispose of it as soon as possible.

Anker says that you shouldn't dispose of the battery "in the trash, general recycling streams (e.g., street-level or curbside recycling bins), or in used battery drop-off boxes commonly found at retail or home improvement stores".

Instead, Anker recommends discarding it at a local electronics disposal facility that accepts recalled devices powered by lithium-ion batteries. That's because recalled batteries need to be treated differently to regular ones due to their higher fire risk.

You might also like

• World's first personal 3D texture printer from Anker just smashed the record for biggest Kickstarter campaign ever – and it's now topped over $46.5 million in pledges
• Disappointed by the Nintendo Switch 2's battery life? I quadrupled it with this cheap accessory
• I tested the Anker 20,000mAh 87W Power Bank and it has plenty of power, but it might be too much for some

• OWC Express 4M2 enclosure offers an alternative route to large, fast external storage
• Thunderbolt 3 on Windows cripples performance to well below the advertised maximum speed
• OWC Express 4M2 SSD slots are PCIe 4.0 x1 only, so individual drive speeds are limited

In a market saturated with expensive high-capacity storage, the OWC Express 4M2 enclosure offers an alternative route to large, fast external storage without immediately breaching the $3000 mark.

At $239.99 for the base configuration, this device cheaper than the TerraMaster D4 SSD and offers a flexible foundation for building what could amount to a 32TB setup when paired with four 8TB NVMe drives.

The company promotes this device as capable of up to 3200MB/s throughput, but real-world performance is highly variable.

Maximum speed requires RAID and careful system configuration

The four M.2 NVMe slots support only PCIe 4.0 x1, which limits individual drive performance to about 1600MB/s.

Reaching peak speeds, therefore, requires RAID configurations and optimal conditions, factors that introduce complexity.

It provides support for RAID 0, 1, 4, 5, and 10, but again, achieving these benefits depends on software licensing, drive quality, and user knowledge.

Users might be drawn by the enclosure’s speed, but should be aware that performance gains require effort and understanding.

Compatibility with USB4 and Thunderbolt standards across macOS and Windows gives the enclosure broad appeal, although actual speed will be gated by the host device.

For example, systems running on older Thunderbolt 3 ports under Windows are capped well below full bandwidth.

While macOS users gain extra features such as booting from RAID arrays, this is limited to systems running at least macOS 11.3.

The Express 4M2 does deliver in terms of construction and thermal management.

Its aircraft-grade aluminum chassis is paired with a dual-fan system that activates only under high thermal load, which ensures quiet operation in most scenarios.

OWC's inclusion of SoftRAID on some models introduces functionality typically reserved for more expensive storage systems.

For those trying to assemble a portable SSD setup or replace their external HDD with something faster, this enclosure makes economic sense, but only just.

However, the cost of filling all four bays with quality 8TB SSDs still pushes the total towards $3000, making this option practical only for those who can supply their own drives or already have SSDs on hand.

You might also like

• These are the fastest SSDs you buy right now
• Take a look at some of the best external hard drives
• Millions of Windows 10 business devices haven't been updated to Windows 11

A fresh selection of new Prime Video movies is on the way to the streaming platform throughout July, delivering a mix of genres, from drama and science-fiction to action and comedy, it’s all here. So whatever you’re in the mood for, there’s plenty to add to your Prime Video watchlist throughout the month. It’s easy to see why Amazon’s streamer continues to be one of the best streaming services around for movie lovers.

In our guide below, we’ve listed every new Prime Video movie that’s arriving in July, along with when you can expect each of them. If you’re only after the best of the bunch, then head straight to our list of the best Prime Video movies available now. Otherwise, read on to find out what’s on its way to Prime Video this July.

New Prime Video movies: July 1

• 2 Fast 2 Furious (2003)
• A Bridge Too Far (1977)
• A Fish Called Wanda (1988)
• Baby Boom (1987)
• Back to School (1986)
• Bandits (2001)
• Blown Away (1994)
• Blue Crush (2002)
• Blue Velvet (1986)
• Colors (1988)
• Creed II (2018)
• Death Warrant (1990)
• Double Impact (1991)
• Dressed to Kill (1980)
• Duck Soup (1933)
• Every Day (2018)
• Fast & Furious 6 (2013)
• Fast Five (2011)
• Fiddler on the Roof (1971)
• Gladiator (2000)
• Hart's War (2002)
• Hidden Figures (2017)
• Hoodlum (1997)
• How to Train Your Dragon 2 (2014)
• I Now Pronounce You Chuck & Larry (2007)
• Into the Blue (2005)
• Into the Blue 2: The Reef (2009)
• It Came From Outer Space (1953)
• K-PAX (2001)
• Leaving Las Vegas (1996)
• Licorice Pizza (2021)
• Little Fockers (2010)
• Little Man (2006)
• Mad Max (1980)
• Meet the Fockers (2004)
• Meet the Parents (2000)
• Mystic Pizza (1988)
• No Country for Old Men (2007)
• No Way Out (1987)
• Only Lovers Left Alive (2013)
• Paths of Glory (1957)
• Rings (2017)
• Robocop (1987)
• Robocop (2014)
• Robocop 2 (1990)
• Robocop 3 (1993)
• Rocky (1976)
• Rocky Balboa (2006)
• Rocky II (1979)
• Rocky III (1982)
• Rocky IV (1985)
• Rocky V (1990)
• Salt (2010)
• Saving Private Ryan (1998)
• Teen Wolf (1985)
• The 40 Year-Old Virgin (2005)
• The Addams Family (2019)
• The Addams Family 2 (2021)
• The Apartment (1960)
• The Bone Collector (1999)
• The Bounty Hunter (2010)
• The Fast and the Furious (2001)
• The Fast and the Furious: Tokyo Drift (2006)
• The Funhouse (1981)
• The Great Train Robbery (1979)
• The Horse Soldiers (1959)
• The House Bunny (2008)
• The Hustle (2019)
• The Informant! (2009)
• The Perfect Storm (2000)
• The Usual Suspects (1995)
• Thunderbolt and Lightfoot (1974)
• What's the Worst That Could Happen? (2001)
• Windtalkers (2002)
• Witness for the Prosecution (1958)
• XXX: The Return of Xander Cage (2017)

New Prime Video movies: July 4

• Game Night (2018)

New Prime Video movies: July 11

• Better Man (2025)

New Prime Video movies: July 15

• Uncharted (2022)

New Prime Video movies: July 19

• Sabotage (2014)

New Prime Video movies: July 25

• Wicked (2024)

New Prime Video movies: July 30

• War of the Worlds (2005)

For more Prime Video-based coverage, check out our guides on The Night Agent season 2, Fallout season 2, Mr and Mrs Smith season 2, and the best Prime Video shows.

Rumors that Apple might make an affordable, multi-colored MacBook based on the A18 Pro chip sparked considerable excitement and ignited a hope in me and others that it could mark the return of the 12-inch MacBook.

Unveiled a decade ago, the gold-finished 2-lb, full-sized keyboard-sporting MacBook was, for its time, an engineering marvel. Apple arguably rewrote its laptop playbook to create the system.

Shortly after launch, I spoke to the marketing head Phil Schiller and Apple's Mac and iPad lead (and now Apple's senior vice president of Hardware Engineering) John Ternus about all Apple did to make the portable wonder.

The pair spread out before me things like a multi-tiered battery and something called the "speaktenna", which was basically a combination of Wi-Fi and Bluetooth antennas and a speaker system.

"We ended up with a group of antenna engineers who know more about speakers than any other antenna engineers and a group of speaker engineers who knew more about antenna design than just about anyone else in the world," Ternus told me.

There was a passion around the product that rivaled that of the Apple Watch, which launched alongside it.

The 12-inch MacBook was also a bit of an odd duck. It was lighter than a MacBook Air, but it was not an Air. It didn't have the same number of ports. In fact, there was just one USB-C port running at USB 3.1 speeds (pokey compared to the Thunderbolt 4 ones you find on today's MacBook airs). Oh, and did I mention that the port handled charging duties, too?

Encrusted with components on both sides, the 12-inch MacBook's motherboard was small enough to fit in the palm of my hand. Everything about the MacBook's components was built in support of its enviable proportions.

At its thickest part, it was, at 0.35cm, thinner than the current MacBook Air. I love the current 13-inch MacBook Air (M4), but I'd be lying if I said I wouldn't enjoy carrying around an even lighter, thinner, and smaller cousin.

It's fair, though, to ask why the market needs such a system now.

First, let's imagine what the MacBook 12-inch 2026 might be. It would have:

• An A18 Pro CPU
• 13-inch LED-backlit Retina display
• 16 GB of memory to support Apple Intelligence
• 128 GB of on-board storage
• 30GB of free iCloud storage
• A fanless design
• A MagSafe Charge port
• A USB-C style Thunderbolt 3 port
• Recycled aluminum enclosure
• A full-sized magic keyboard including Touch ID
• A 4-inch Force Touch Trackpad

Design-wise, the 12-inch MacBook A18 Pro would align closely with the MacBook Air line. No more wedge, instead two flat panels squeezed together into a 0.35cm-thick slab.

Granted that everything above is guesswork, but I believe that configuration would fit neatly into a $599 package (maybe even a $499 one).

Nothing here is new, and the A18 Pro is plenty powerful and efficient to run such a system.

The benefit, obviously, is an affordable, yet nearly full-sized portable that is a complete system. I am well aware you can buy an M4 Mac Mini for $599, but you still need to buy a mouse, keyboard, and screen. This, by contrast, would be the full Apple MacBook package at, finally, an affordable price.

You might have also noticed the rather paltry base storage. That's to help keep costs down. It's buttressed, though, by something Apple desperately needs to do: offer more versatile and forgiving iCloud storage options.

The usual 5GB of free storage is not enough, and I think the extra 30GB will offset the limited local storage, moving those who are on the fence about the 12-inch MacBook into the must-buy territory.

Give them what they want

If Apple balks at reintroducing the confusing "MacBook" name, especially when it's smaller and lighter than the MacBook Air, they could call it the MacBook Air LT (for light) or MacBook Air A (for its A-series chip).

One need only look at Walmart for evidence that consumers want such a system. It's been selling the old-school design MacBook Air M1 for $699 and now $649 for years. People are desperate for a truly affordable Mac, but they are probably tired of the growing performance compromises tied to the aging M1 chip.

The A18 Pro will sing in a tiny 12-inch laptop, and the system's incredibly small proportions will make it a hit with those who thought they might go for a lightweight iPad with a Magic Keyboard Folio, mainly because they thought it wouldn't weigh down their backpack.

Apple has learned, thanks to Apple Silicon, so much about building lightweight and performant systems that it makes sense to extend the MacBook idea in new and maybe unexpected directions.

A revived 12-inch MacBook would sell like hotcakes and pave the way for more fresh ideas, like a 12-inch MacBook Air running an M3 chip. That one could sell for $699.

You might also like

• Apple could launch a cut-price MacBook powered by an iPhone chip, new report claims – here’s why that would be a massive hit
• Constantly dismissing notifications on your Apple Watch? You’re going to love Apple’s watchOS 26 latest gesture
• 11 great uses for a smart plug: upgrade all the 'dumb' devices in your home for an easier life

Are modern smartphone designs boring? We polled 1,500 readers for their thoughts on that very question earlier this year, and a whopping 78% of respondents answered in the affirmative.

The smartphone industry was once awash with weird and wonderful devices (remember the LG Wing?), but these days, manufacturers are – quite literally – much more straight-edged in their approach to design.

The exception to the rule is Nothing, which has stuck by its mission to “make tech fun again” since being founded by former OnePlus chief Carl Pei in 2021. Nothing phones are designed to stand out from the crowd, and the company’s latest handset, the long-awaited Nothing Phone 3, fulfils that brief and then some.

Nothing’s “first true flagship” is a truly weird-looking thing. Imagine if a kindergartner took an old-school sliding block puzzle game, dipped it in PVA glue, and accessorized it with three black sewing buttons. That’s the Nothing Phone 3. It’s an outlandish creation with specs to rival some of the best phones – and I think it could be the most interesting flagship launch of the year.

So, what’s new versus the well-received (but decidedly un-flagship) Nothing Phone 2? For starters, linear Glyph lights are out, and a new Glyph Matrix interface is in.

This interactive screen sits in the top-right corner of the Phone 3’s 6.67-inch rear panel and is capable of displaying several different types of content – from person-specific notifications to ‘Glyph Toys’ (read: party games) like Spin the Bottle and Rock, Paper, Scissors. Heck, Glyph Matrix can even double as a pixelated viewfinder for selfies, which is a nice retro touch for those who miss the Game Boy Camera.

Image 1 of 3

The new Glyph Matrix interface on the Nothing Phone 3 (Image credit: Future)Image 2 of 3

The Magic 8 Ball Glyph Toy (Image credit: Future)Image 3 of 3

The Magic 8 Ball Glyph Toy (Image credit: Future)

As for other big design changes, the Phone 3 still has that classic modular Nothing look. But its camera lenses sit independently of a physical module, rather than being positioned in a circular or pill-shaped casing, as on the Phone 3a and Phone 3a Pro, respectively.

I can see this new lens arrangement being the most controversial aspect of the Phone 3’s design, but I’m personally a big fan of the abstract, asymmetrical vibe it brings to the device as a whole. Previous Nothing phones looked unique, sure, but not to this extent.

(Image credit: Future)

Speaking of cameras, there’s a brand new one. The Phone 3 gets a 50-megapixel main lens, a 50MP ultra-wide lens, and an all-new 50MP telephoto lens with 3x optical zoom. Of course, we can’t yet say how good these cameras are without having tested them properly (stay tuned for our full review), but on paper, at least, they’re flagship-level snappers.

You’ll also get a 50MP selfie camera on the front, which marks a step up from the 32MP equivalent on the Nothing Phone 2. Oh, and the little red dot on the back of the Phone 3? That’ll light up when you’re recording with any one of its three rear cameras.

The Nothing Phone 3 running Nothing OS 3.5 atop Android 15 (Image credit: Future)

Under the hood, the Phone 3 is powered by the new Qualcomm Snapdragon 8s Gen 4 chipset. I’ve already written about why I don’t think the lack of a Snapdragon 8 Elite is a big deal in a separate piece, but we won’t know for sure until we’ve benchmarked the Phone 3 against similarly priced rivals.

Nothing, for its part, is confident in the performance credentials of its “first true flagship,” and having had a quick play with the Phone 3 at its London launch event, I’m convinced that the 8s Gen 4 will be speedy enough to support the software demands of Nothing OS 3.5 (and, eventually, Nothing OS 4.0).

The Phone 3’s chipset is paired with a 5,150mAh silicon-carbon battery, which is a technology you still won’t find in many of the best Android phones. You’ll get 65W fast charging, too – another notable upgrade over the Phone 2’s 45W charging speed cap.

Other features of Nothing’s newest flagship include an IP68 dust and water resistance rating, Corning Gorilla Glass protection, and a new Essential Search function to go alongside the physical Essential Key, which was introduced on the Phone 3a Pro.

Again, I only handled the phone for a few minutes at Nothing’s launch event, so I haven’t yet taken these new tools for a proper spin. But the bottom line is this: the Phone 3 is unlike any other device on the market right now, and I sincerely hope that its performance lives up to Nothing’s “flagship” billing. If the Snapdragon 8s Gen 4 can compete with Google’s Tensor G5, at the very least, then the Phone 3 could well be the most interesting phone of the year.

The device starts at $799 / £799 for the model with 12GB of RAM and 256GB of storage, with that price rising to $899 / £899 for 16GB and 512GB, respectively. Those figures put the Phone 3 in league with the iPhone 16, Samsung Galaxy S25, and Google Pixel 9, so stay tuned for our verdict on whether Nothing’s latest phone can compete with such big-name rivals.

You might also like

• The best Android phones 2025: top performing and most affordable
• The best Android tablets 2025: powerful tablets for every budget
• The best phone 2025: we've tested the top phones you can buy in the US

• WinRAR flaw let crafted archives drop files outside target folder, including into Windows Startup
• New version 7.12 addresses critical path and HTML vulnerabilities
• Windows users urged to update WinRAR for improved file safety

Iconic file archiving tool WinRAR has received a security update addressing a serious flaw that could let attackers run arbitrary code on affected systems.

The vulnerability, tracked as CVE-2025-6218, was identified in the way WinRAR handles file paths within archives.

It was discovered by a researcher known as whs3-detonator, working with Trend Micro’s Zero Day Initiative.

Patch now

The issue exists in Windows versions of WinRAR, where a specially crafted archive can exploit path traversal during file extraction.

If a user opens such a file or visits a malicious site, the exploit can allow files to be placed in unintended directories, including sensitive ones like the Windows Startup folder.

This could cause malicious software to run automatically when the system boots.

RARLAB, the developer of WinRAR, has released version 7.12 to address this flaw.

The vulnerability does not affect versions of RAR or UnRAR for Unix or Android. Users are urged to update as soon as possible to reduce the risk of exploitation.

To stay protected from threats like this, it’s important to use the best antivirus software, reliable malware removal tools, and strong endpoint protection. Even well-known tools can have flaws, so running trusted security software and keeping all apps current helps reduce the risk of malware slipping through unnoticed.

The new WinRAR update also fixes an unrelated issue involving the “Generate Report” feature. In older versions, file names in generated HTML reports weren’t sanitized properly, which allowed basic HTML injection. That has now been corrected.

In addition to the security fixes, WinRAR 7.12 now tests recovery volumes during archive testing, giving users better confirmation that backup files are intact. It also preserves precise nanosecond timestamps when modifying Unix files on Windows.

You might also like

• We've also 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