TechRadar News

The Asus ROG Xbox Ally X20 feels like the PC gaming handheld of 2026. An enhanced version of the original Xbox-branded Ally X, this is shaping up to be one of the best-looking, ergonomically fantastic, and brilliantly performing handheld games consoles — and I went hands-on with it at Gamescom last month and liked it. A lot.

Coming to it as someone who hasn’t truly jumped in with both feet to PC gaming handhelds — basically none of them play nice with my hands due to being bulky and heavy — even a short time with the newest, flashiest version of Asus’ Xbox handhelds has left a permanent mark. One that might make me a PC handheld convert.

Much will be made of its price tag and purchase options — either $1,299 / £1,299 on its own, or $2,299.99 / £2,299.99 as a bundle with the new ROG XREAL R1 glasses — but after experiencing all its enhancements and features, and seeing how games run on it, it really could be a hit.

Looks and feels the part

(Image credit: Future)

Even though I know that this is a similar handheld to its predecessor, this thing is a joy to hold and use from a design perspective.

The overarching form is the same as the original Ally X, but the new translucent finish is a thing of beauty. It gives it an almost see-under-the-hood kind of vibe, and does a great job of making it stand out from its largely black-and-white competitors.

The gold accents are even more fantastic in real life and up close than in pictures, too, with the rings around each thumbstick and the shine of the four small buttons on the front, as well as the ‘ROG 20th Anniversary’ plate, and the stripe down the back really popping out of the otherwise black aesthetic.

(Image credit: Future)

That translucent back also has an incredibly grippy feel — you won’t need to add an external case here. As someone who always has to worry about that with gamepads and phones, this was extremely welcome; the rubberised, tactile wings of the controller really are conducive to maintaining a solid grip.

The inputs remain great too, with rounded face buttons, lovely triggers and bumpers that are all a joy to interact with, as well as excellent Tunnel Magnetoresistance (TMR) thumbsticks that offer a tangible upgrade over the analogue sticks of its predecessor. Though, I’d have liked to have seen some microswitch button tech in the X20 as a similar upgrade as well.

The star of the show, though, is, of course, that slightly bigger, but much better, screen. Offering an increase of 0.4 inches, but a huge leap up to OLED in panel tech, the new screen is vibrant and has colors that really pop and contrast. The boost to color gamut and HDR it offers makes for a genuine visual treat, and you know you’ll be covered in all situations with a brightness of up to 1400 nits to tap into.

But has the guts to match

(Image credit: Future)

The performance remains excellent as it was on the X20; games might never have looked better on a handheld screen.

The OLED panel is a joy to use, with fantastic colors and contrast, and the extra bump in the VRR to FreeSync Premium Pro is also welcome and a great feature. As a result, Forza Horizon 5 looks comfortably better, by a strong margin, compared to the handheld I am most used to, my PlayStation Portal, in sheer image quality and vibrancy, while getting comfortably north of 60 frames per second.

Using the TMR thumbsticks was great too: every move feels much more responsive, smooth, and instant compared to my Portal or a Steam Deck, and the extra durability the sticks offer over analog ones is superb too.

The Asus ROG Ally X20 is a fantastic bit of kit, and offers a compelling route into handheld PC gaming — perhaps one of the most complete offerings — even for me, as someone who has yet to jump in and can’t ever find the right device for the job.

So far it might be the PC gaming handheld I’ll consider buying most if I commit to going mobile. It’s not a transformation of the PC handheld experience that the ROG Xbox Ally X offers, but it’s an excellent refinement.

Twitter is not back, and we'll surely never see the likes of it again. In its place, we have X, and now, Tweet.app. You can be forgiven for the confusion. Many who still use X think their posts are tweets, but the still-pay-to-play platform from Bluebird, Inc. is now called Tweet.

It's a tale far too rich to contain in, say, a classic 140-character tweet, but I'll attempt to unpack it here, with insights from Bluebird, Inc. President and General Counsel Stephen Coates.

"I call it the 'Tweenix,' " Coates tells me, after I noted that Tweet.app's new blue logo looks a little like a phoenix. The imagery is apt. It signifies classic Twitter rising from the ashes, but also Bluebird's initial success and then failure to revive 'Twitter', before ultimately launching Tweet.app a few weeks ago.

Two years ago, Bluebird began the skunkworks project to revive Twitter, a brand Elon Musk effectively killed when he bought Twitter in 2022 and then in 2023 rebranded it to X, leaving vestiges of the old 'Twitter' scattered throughout the services, a fact that played a crucial role in Bluebird's ability to revive the brand.

Coates told me that after his company tried to file for a cancellation of the 'Tweet' and 'Twitter' trademarks in December of last year, X quickly filed a federal injunction. In the meantime, Bluebird launched Twitter.now. Months later, the judge ruled that because X still had "formerly Twitter" on the Apple App Store listing, it had brand rights. Coates called the ruling surprising, but Bluebird quickly pivoted and rebranded to Tweet.

The company, he told me, has no plans to go after the iconic Twitter bird logo, even though it's not well protected as a brand by X. Hence, the Tweenix.

Tweet, though, still seems like a risky name. Coates told me, "We're optimistic... It seems highly likely we’ll prevail on 'Tweet'."

Bluebird, Inc. President Stephen J. Coates (Image credit: Bluebird, Inc.)

That remains to be seen. When you sign up for Tweet, you must acknowledge that you know it's not affiliated with X. Search Tweet in the App Store, and the first result is still X. However, Tweet.app operating relatively under the radar might help it avoid further scrutiny from X and Elon Musk.

It has, by Coates' measure, just about 8,000 users. "[That's] pretty good, considering it's paid," Coates told me. The platform is so small that you could conceivably read all the tweets — I'm not sure that's a selling point, but Coates said there are "already community and inside jokes."

The company is using the income from those users to pay its small team of 10 employees and contractors as it builds the platform and tweaks the moderation filters, something Coates considers of critical importance.

Tweet.app, which does resemble old-school Twitter, is currently using AI to scale its moderation, including its still-in-beta Vera tool, which measures content for misinformation and applies a score. Eventually, Tweet.app members will have a dial connected to those ratings to automatically filter out their desired level of misinformation.

That's right, Tweet.app won't be in the business of rooting out all fake news. The creed will be, "Freedom of speech but not freedom of reach," said Coates.

FutureFutureFuture

Where Twitter, under former CEO Dick Costolo, whom Coates worked under, struggled with how to be a town square when there was one guy shouting obscenities in the corner, Tweet.com will let the guy shout, but also let users hide the obscenities they don't want to see in their feeds.

If it works, all of it will be powered by this AI. "I call it AI for good. We're not creating tools that will create AI outputs," said Coates, who has worked as counsel for Meta, Reddit, and TikTok.

None of this, of course, ensures success. Classic social media, including Twitter, was built on viral moments, like the Arab Spring, the killing of Osama Bin Laden, and selfies that broke the internet.

I call it AI for good. We're not creating tools that will create AI outputs

Stephen Coates

Coates acknowledged the power of these "inflection" moments, but reminded me that you can't predict them. Tweet.app is happy to grow slowly. Coates knows that paying members will be more deeply engaged than people who simply hop on the platform for free. "At some point will it be open to everyone? I certainly hope so," he told me.

I asked Coates why we need this or any other new social media. He told me that he sees room for reinvention, with space for social media that gives users control. "We're empowering them to control the algorithm," which is an idea that well-established social media platforms have only recently approached.

As the other upstart micro-blogging social media platform, Bluesky, moves away from the town square concept, Coates sees an opportunity to build an alternative to the only other big open platform: X.

"There's room for more. I think people will come for the nostalgia of Twitter and tweeting and stay for the control and experience. It’s a grand experiment."

• CISA added three Linux kernel flaws (CVE‑2025‑39682, CVE‑2026‑53266, CVE‑2025‑39964) to KEV catalog
• Red Hat confirmed active exploitation; agencies given rare three‑day patch deadline expiring Sept 21, 2026
• Bugs enable DoS, privilege escalation, or data corruption; patches available, limited mitigations for two flaws

The US Cybersecurity and Infrastructure Security Agency (CISA) has added three Linux flaws to its Known Exploited Vulnerabilities (KEV) catalog, signaling abuse in the wild and giving government agencies a deadline to patch or stop using the flawed product entirely.

The three bugs in question are tracked as CVE-2025-39682 (severity score 9.8/10 - critical), CVE-2026-53266 (severity score 8.8/10 - high), and CVE-2025-39964 (severity score 7.8/10 - high). All three have already been patched in the Linux kernel. CVE-2025-39682 was fixed in stable releases 6.1.149, 6.6.103, 6.12.44 and 6.16.4, while CVE-2025-39964 was fixed in 5.10.245, 5.15.194, 6.1.154, 6.6.108, 6.12.49 and 6.16.9. CVE-2026-53266 has been fixed upstream and backported to supported stable/distribution kernel branches, including 5.10.259, 6.1.176 and 6.12.94.

The first issue is an improper check for unusual or exceptional conditions vulnerability in the TLS receive patch which could allow unauthenticated threat actors to launch memory disclosure or denial-of-service (DoS) attacks. The second one (CVE-2026-53266) is an out-of-bounds write vulnerability in the ebtables Source Network Address Translation (SNAT) Address Resolution Protocol (ARP) rewrite patch which allows local attackers to escalate privileges or mount DoS attacks.

The last one (CVE-2025-39964) is a race condition flaw that allows concurrent writes to the same AF_ALG socket, which allows local malicious actors to crash the system or corrupt cryptographic operation results, leading to data integrity issues and possible DoS states.

Attacks in the wild

Red Hat acknowledged that all three are being abused in real-life attacks. “This CVE is high risk and there are known public exploits leveraging this vulnerability. Address this vulnerability with high priority,” it said in all three advisories.

However, there are no details as to who is currently using these exploits, against whom, and to what cause. There are currently no separate reports of cyberattacks referencing any of the abovementioned vulnerabilities.

However, CISA still reacted. All three flaws were added on September 18, 2026, and all three have a small three-day deadline for patching that expires on September 21. Usually, CISA would grant Federal Civilian Executive Branch (FCEB) agencies a three-week deadline to patch up, with just exceptionally dangerous flaws getting a shorter window. That being said, these flaws are likely extremely dangerous.

In theory...

In a hypothetical scenario, a threat actor could target a Linux system using kernel TLS (kTLS) by sending a specially-crafted TLS record that triggers CVE-2025-39682 and causing either a crash, or even arbitrary code execution. Attackers that already have a low privilege foothold on the target endpoint could use CVE_2025-39964 to escalate privileges, while on systems using the affected bridge/netfilter configuration, CVE-2026-53266 could be used for privilege escalation, as well.

According to the Red Hat advisory, there is a chance that CVE-2025-39682 is remotely triggerable, but only when the system is using the affected kTLS receive path. The other two flaws are exclusively local vulnerabilities.

Besides fixes, two out of the three flaws also have possible mitigations. For the improper check one, users should prevent module tls from being loaded. For the out-of-bounds write one, users are advised to disable ARP hardware address rewriting in ebtables SNAT rules, or remove ebtables SNAT rules that operate on ARP traffic on bridge interfaces. The final vulnerability, currently does not have a working mitigation, and the only way to stay secure is to apply the provided patches.

Linux kernel vulnerabilities are generally considered serious, but the severity still depends on the flaw and how the affected kernel is deployed. Earlier this year, security researchers disclosed four local privilege escalation flaws, called DirtyAH6 (CVE-2026-80844), TUNderflow (CVE-2026-81000), PPPoEject (CVE-2026-68121), and DiagSpill (CVE-2026-74469).

Via The Hacker News

For Stuart Fails to Save the Universe viewers who have been waiting for things to feel exactly like The Big Bang Theory... your moment has arrived.

Sure, we've had a whole host of cameos from the original show (though three notable faces have yet to turn up), but zipping from unhinged universe to unhinged universe has been a world away from the science sitcom we know and love.

But what if I said that there's a key reason why this week's episode could be straight out of 2007? And when does Stuart Fails to Save the Universe episode 9 arrive on HBO Max?

What time can I watch Stuart Fails to Save the Universe episode 10 on HBO Max?

For US viewers, Stuart Fails to Save the Universe episode 10 will drop on Thursday, September 24 at 6pm PT/9pm ET.

Internationally, you're looking out for these times:

• US – Thursday, September 24 at 6pm PT / 9pm ET
• Canada – Thursday, September 24 at 6pm PT / 9pm ET
• UK – Friday, September 25 at 2am BST
• India – Friday, September 25 at 6:30am IST
• Singapore – Friday, September 25 at 9am SGT
• Australia – Friday, September 25 at 11am AEST
• New Zealand – Friday, September 25 at 1pm NZST

When do new episodes of Stuart Fails to Save the Universe come out?

(Image credit: HBO Max)

New episodes of Stuart Fails to Save the Universe will make landfall every Thursday in the US and on Fridays everywhere else. Here are the all-important dates you need to know about:

• Episode 1: out now
• Episode 2: out now
• Episode 3: out now
• Episode 4: out now
• Episode 5: out now
• Episode 6: out now
• Episode 7: out now
• Episode 8: out now
• Episode 9: out now
• Episode 10: September 24

• HKUST (Guangzhou) and HK PolyU researchers' InjectEave beams a radio carrier at devices so their own circuits leak analog audio
• The leaked audio has already been decrypted by the device itself, making encryption offer no protection against such an approach
• The 30m headline needed a pricey amplifier pushing output to 10 W, while standard ranges of 1 to 6m were measured by detecting a test tone

Researchers at the Hong Kong University of Science and Technology and the Hong Kong Polytechnic University have shown that everyday headphones, a desk phone, and a handful of smart-home gadgets can broadcast what they are doing simply by using a radio signal.

The technique, called InjectEave, was presented at USENIX Security 2026 in Baltimore, and multiple outlets have since covered the researcher's claim that the attack "can recover headphone audio from up to 30 meters away, including through walls".

The claim may be accurate, but it has limitations, including the need for specialized equipment that is often very noticeable in most settings.

An impressive technical show with plenty of limitations in tow

Conventional electromagnetic eavesdropping waits for a device to leak. That works poorly for audio, because speech sits below 20 kHz while a device's wiring radiates efficiently only at megahertz or gigahertz frequencies.

InjectEave chooses to close the gap by transmitting a carrier signal at the target. According to the paper, nonlinear components such as amplifiers, analog-to-digital converters, switching MOSFETs, and power converters mix the secret signal onto that carrier, and the device's own traces and cables radiate the result back to a receiver.

Because the leak happens in the analog path, after audio has been decoded, the project page states that "InjectEave is immune to digital defenses such as encryption, masking, and randomization." Encryption in any form on a wireless headset is irrelevant, since the attack directly targets the signal driving the speaker, not the radio link.

The team demonstrating this used a USRP B210 software-defined radio, two antennas, a Siglent spectrum analyzer, and a laptop to set up their proof of concept, which could, as they noted, "eavesdrop on the majority of these devices from over 2m away and through walls, with a maximum distance of 30m for recovering intelligible headphone audio".

This isn't the first time the technique has been used, even as the researchers have built on it considerably; the idea originates from "The Thing," a Soviet bug given as a gift to the US ambassador in Moscow in 1945, which was passively powered remotely as a listening device.

InjectEve, however, does not involve planting anything; instead, it leverages existing work on interference-induced leakage and impedance-based backscatter, and distinguishes itself by recovering coarse digital data rather than continuous analog waveforms, allowing users to 'listen in' without added steps.

The same approach applies to wired headphones, which are also prone to leaking microphone audio; researchers concluded that their sound cards were the primary source of the leaks, which were being sent back over very short distances.

The threat is somewhat tempered by the fact that it can capture what users hear but, with wireless headphones, cannot hear what the user is saying. The attacker also has to transmit continuously and use a high-powered (10W) transmitter to reach the full 30-meter range, even though capturing and rendering speech is harder than the tones the test used to prove the approach worked.

The attacker also needs to know the target model and profile a matching unit first, although they managed to get a profile to transfer cleanly across three identical UGreen headsets, suggesting this might be easier than one would assume.

The researchers say they reported the findings to the affected manufacturers but, "as we have not yet received a response," withheld the table's frequencies and stripped injection control logic from their released code. One shouldn't be too hopeful, however, as no software update can fix an analog leak, and it is relatively limited in practical abuse cases.

Looking for a different day?

A new Quordle puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: Quordle hints and answers for Monday, September 21 (game #1701).

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,500 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today — or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc's Wordle today column covers the original viral word game.

SPOILER WARNING: Information about Quordle today is below, so don't read on if you don't want to know the answers.

Quordle today (game #1702) — hint #1 — VowelsHow many different vowels are in Quordle today?

• The number of different vowels in Quordle today is 3*.

* Note that by vowel we mean the five standard vowels (A, E, I, O, U), not Y (which is sometimes counted as a vowel too).

Quordle today (game #1702) — hint #2 — repeated lettersDo any of today's Quordle answers contain repeated letters?

• The number of Quordle answers containing a repeated letter today is 2.

Quordle today (game #1702) — hint #3 — uncommon lettersDo the letters Q, Z, X or J appear in Quordle today?

• No. None of Q, Z, X or J appear among today's Quordle answers.

Quordle today (game #1702) — hint #4 — starting letters (1)Do any of today's Quordle puzzles start with the same letter?

• The number of today's Quordle answers starting with the same letter is 0.

If you just want to know the answers at this stage, simply scroll down. If you're not ready yet then here's one more clue to make things a lot easier:

Quordle today (game #1702) — hint #5 — starting letters (2)What letters do today's Quordle answers start with?

• C

• L

• D

• P

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

Quordle today (game #1702) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle, game #1702, are…

• COMFY
• LOBBY
• DRESS
• PANEL

After making a speedy start I ground to a halt trying to get my last word.

Fortunately, it only took me two guesses to find COMFY, but I wasted lots of time thinking I was looking for another double-letter word.

Daily Sequence today (game #1702) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle Daily Sequence, game #1702, are…

• SPOOL
• LUPUS
• SHELL
• HAIRY

Quordle answers: The past 20

• Quordle #1701, Monday, 21 September: WAXEN, HIPPO, PETAL, LEASH
• Quordle #1700, Sunday, 20 September: DERBY, LIMBO, GAWKY, REIGN
• Quordle #1699, Saturday, 19 September: SCOPE, HEIST, PULSE, CHAMP
• Quordle #1698, Friday, 18 September: MUCUS, MIRTH, APPLE, TILDE
• Quordle #1697, Thursday, 17 September: CRIER, CHIRP, NEIGH, FREER
• Quordle #1696, Wednesday, 16 September: SPEAR, PRICE, BLUNT, FISHY
• Quordle #1695, Tuesday, 15 September: SHUSH, LITHE, DELTA, BUGLE
• Quordle #1694, Monday, 14 September: ZEBRA, LEGAL, RATIO, TROLL
• Quordle #1693, Sunday, 13 September: LUCID, SPRAY, MANGE, EMAIL
• Quordle #1692, Saturday, 12 September: SUPER, BRUSH, RESET, SOWER
• Quordle #1691, Friday, 11 September: PLUSH, PAINT, RIVAL, AFOUL
• Quordle #1690, Thursday, 10 September: AGENT, EXILE, TITAN, PUREE
• Quordle #1689, Wednesday, 9 September: HAPPY, GONER, SMACK, MOTIF
• Quordle #1688, Tuesday, 8 September: FLUNG, DIARY, HALVE, BERTH
• Quordle #1687, Monday, 7 September: VISIT, CLANK, FEVER, OCEAN
• Quordle #1686, Sunday, 6 September: PILOT, HONOR, BOWEL, TAUNT
• Quordle #1685, Saturday, 5 September: GNASH, GEEKY, WHELP, PUTTY
• Quordle #1684, Friday, 4 September: QUICK, BEEFY, TRUCK, SNAIL
• Quordle #1683, Thursday, 3 September: SIGMA, ESSAY, MERRY, LLAMA
• Quordle #1682, Wednesday, 2 September: INCUR, FELLA, TITAN, FROZE

Looking for a different day?

A new NYT Connections puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: NYT Connections hints and answers for Monday, September 21 (game #1198).

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

NYT Connections today (game #1199) - today's words

(Image credit: New York Times)

Today's NYT Connections words are…

• THIS
• THAT
• THE
• OTHER
• CAFE
• PATRON
• AGREE
• ACCENT
• TOUCH
• THOSE
• ROSE
• NEUTRAL
• DETAIL
• DISAGREE
• THESE
• FLOURISH

NYT Connections today (game #1199) - hint #1 - group hints

What are some clues for today's NYT Connections groups?

• YELLOW: Determiners
• GREEN: Adornment
• BLUE: A quiz with choices
• PURPLE: Fancy drinks

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

NYT Connections today (game #1199) - hint #2 - group answers

What are the answers for today's NYT Connections groups?

• YELLOW: DEMONSTRATIVES
• GREEN: EMBELLISHMENT
• BLUE: OPTIONS ON A MULTIPLE CHOICE SURVEY
• PURPLE: BEVERAGES WITH ACUTE ACCENTS REMOVED

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Connections today (game #1199) - the answers

(Image credit: New York Times)

The answers to today's Connections, game #1199, are…

• YELLOW: DEMONSTRATIVES: THAT, THESE, THIS, THOSE
• GREEN: EMBELLISHMENT: ACCENT, DETAIL, FLOURISH, TOUCH
• BLUE: OPTIONS ON A MULTIPLE CHOICE SURVEY: AGREE, DISAGREE, NEUTRAL, OTHER
• PURPLE: BEVERAGES WITH ACUTE ACCENTS REMOVED: CAFE, PATRON, ROSE, THE

• My rating: Easy
• My score: Perfect

Connections loves to hide a clue to another group on one of the tiles and today it was ACCENT.

This tile featured in the green group for EMBELLISHMENT but may also have tipped some crafty players off to the fact that we were searching for a group featuring accents or, as it happened, BEVERAGES WITH ACUTE ACCENTS REMOVED.

Before linking the fancy drinks I took a bit of a risk with the yellow group, having to choose between THE and THIS. Had I paid attention more at school I may have found this task easier — demonstratively easier, even. Yet somehow I scraped through without a single error.

Yesterday's NYT Connections answers (Monday, September 21, 2026, game #1198)

• YELLOW: EMAIL BUTTONS: COMPOSE, DRAFTS, SENT, SPAM
• GREEN: WAYS TO ATTRACT FISH: CHUM, FLY, SPINNER, WORM
• BLUE: CHESS VERBS: CAPTURE, CASTLE, CHECK, MATE
• PURPLE: STARTS OF COMPOSERS' NAMES: BEET, BRAH, CHOP, VIVA

What is NYT Connections?

NYT Connections is one of several increasingly popular word games made by the New York Times. It challenges you to find groups of four items that share something in common, and each group has a different difficulty level: green is easy, yellow a little harder, blue often quite tough and purple usually very difficult.

On the plus side, you don't technically need to solve the final one, as you'll be able to answer that one by a process of elimination. What's more, you can make up to four mistakes, which gives you a little bit of breathing room.

It's a little more involved than something like Wordle, however, and there are plenty of opportunities for the game to trip you up with tricks. For instance, watch out for homophones and other word games that could disguise the answers.

It's playable for free via the NYT Games site on desktop or mobile.

Looking for a different day?

A new NYT Strands puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: NYT Strands hints and answers for Monday, September 21 (game #932).

Strands is the NYT's latest word game after the likes of Wordle, Spelling Bee and Connections – and it's great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc's Wordle today page for the original viral word game.

SPOILER WARNING: Information about NYT Strands today is below, so don't read on if you don't want to know the answers.

NYT Strands today (game #933) - hint #1 - today's themeWhat is the theme of today's NYT Strands?

• Today's NYT Strands theme is… Day and night shift

NYT Strands today (game #933) - hint #2 - clue words

Play any of these words to unlock the in-game hints system.

• FAIL
• SAVERS
• TWIN
• CHAIN
• VOTES
• SOIL

NYT Strands today (game #933) - hint #3 - spangram lettersHow many letters are in today's spangram?

• Spangram has 11 letters

NYT Strands today (game #933) - hint #4 - spangram positionWhat are two sides of the board that today's spangram touches?

• First side: top, 1st column

• Last side: bottom, 4th column

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Strands today (game #933) - the answers

(Image credit: New York Times)

The answers to today's Strands, game #933, are…

• SWITCH
• CHANGEOVER
• TRANSITION
• PROGRESSION
• SPANGRAM: FALLEQUINOX

• My rating: Hard
• My score: 1 hint

Even as I was collecting the words I still thought that the theme was about work and the moment when the day shift and the night shift swap over.

It wasn’t until I got the spangram — I had been ignoring the game’s two rare letters until then — that the penny finally dropped and I understood that the words were connected to what AI tells me is “the precise astronomical moment when the center of the Sun crosses Earth's equator, marking the official astronomical start of autumn”.

Every day’s a school day with Strands!

Yesterday's NYT Strands answers (Monday, September 21, game #932)

• NEON
• LYCRA
• GRUNGE
• SCRUNCHIES
• TRACKSUITS
• SPANGRAM: NINETIESTREND

What is NYT Strands?

Strands is the NYT's not-so-new-any-more word game, following Wordle and Connections. It's now a fully fledged member of the NYT's games stable that has been running for a year and which can be played on the NYT Games site on desktop or mobile.

I've got a full guide to how to play NYT Strands, complete with tips for solving it, so check that out if you're struggling to beat it each day.

• Hackers are using blockchains to keep malware instructions available after servers disappear
• AI is making blockchain-based malware infrastructure easier for less experienced hackers
• Blockchain traffic is difficult to block without disrupting legitimate cryptocurrency services worldwide

Hackers are increasingly hiding malware instructions inside public blockchains, creating communication channels that can survive the removal of conventional infrastructure.

New figures from Chainalysis claim malicious blockchain activity increased 440%, with daily entries rising from 2.06 to 11.1 after newer AI systems emerged.

The technique gives attackers another way to maintain communication with compromised computers without relying entirely on conventional servers controlled by hosting providers.

Blockchain networks become malware dead drops

Blockchain dead drops use transaction data or smart contracts as lookup points, allowing infected computers to retrieve commands, addresses, or configuration information.

Because blockchain records are distributed across networks, removing a conventional server does not erase information already stored on the ledger.

A North Korean-linked operation associated with UNC5342 uses TRON and Aptos as alternate routes before retrieving encrypted instructions through the BNB Chain.

Its malware can check one network, switch to another when necessary, and retrieve updated addresses without receiving another malware package.

Iranian actors suspected of links to the country's intelligence ministry have embedded encoded routing information inside Bitcoin transactions used for malware retrieval.

Russian-speaking cybercriminals have also commercialized the technique, using Polygon contracts to provide blockchain-backed infrastructure for malware campaigns operated by different customers.

One related operator controls more than 50 BNB Chain resolver contracts while also conducting activity involving fraudulent tokens and clipboard-monitoring malware.

These operations show how blockchain records can function as persistent lookup infrastructure rather than merely serving their conventional financial and transactional purposes.

AI lowers the technical barrier

Chainalysis said the sharp increase in this malicious activity followed the arrival of high-capacity Chinese open models, which placed fewer restrictions on malware development requests.

Before those systems appeared, building reliable blockchain-based malware infrastructure required expertise across malicious software, cryptocurrency networks, and distributed communication systems.

AI tools can reduce that knowledge barrier by helping less experienced operators understand unfamiliar technologies and produce components needed for blockchain communication.

In the second quarter of 2026, state-linked groups accounted for roughly two-thirds of newly observed activity.

Those groups also represent about half of overall observed activity, indicating that blockchain-based malware infrastructure extends beyond conventional cybercriminal operations.

Defenders face difficulties because blocking blockchain traffic could also disrupt legitimate wallets, decentralized applications, exchanges, and decentralized finance services used worldwide.

Attackers can further complicate disruption by operating their own blockchain nodes, reducing dependence on external providers that defenders might otherwise pressure or disable.

Some operators have hidden server addresses inside wallet identifiers without usable private keys, then used zero-value transfers to trigger malware retrieval.

Those transactions leave public records that investigators can examine, potentially providing useful clues even when attackers attempt to conceal their infrastructure.

"While the exploitation of blockchain by state-linked organizations such as North Korea is becoming more sophisticated, on-chain records left by attackers can actually serve as important clues to track them," said Kwon Jun-hyeok, General Manager of Chainalysis Korea.

"Tracking these traces and identifying attackers and related infrastructure through blockchain intelligence will become increasingly important in responding to new cyber threats."

• Huawei’s Ascend 960 roadmap depends on scale, memory, and faster interconnects
• UnifiedBus connects thousands of processors while reducing communication bottlenecks inside AI systems
• The Atlas 960 roadmap reaches 15,488 processors with 34 PB/s bandwidth

Huawei is expanding its Ascend platform with larger systems designed to handle AI models containing up to 10 trillion parameters.

The company is advancing UnifiedBus, an interconnect architecture intended to connect processors, memory, and storage with lower communication overhead.

That approach gives Huawei another way to compensate for hardware limitations that have constrained individual Ascend accelerators against Nvidia's products.

Huawei expands the Ascend 960 system

Huawei's Atlas 960 SuperPoD roadmap calls for systems containing as many as 15,488 Ascend 960 processors, up from its earlier 8,192 processor configuration.

The company says the system can deliver 30 EFLOPS of FP8 computing and 60 EFLOPS using FP4, alongside 4,460 TB of memory capacity.

Its interconnect bandwidth is specified at 34 PB/s, while Huawei expects training performance to reach 15.9 million tokens per second.

Huawei has also accelerated development of the Ascend 960 family, with the 960DT scheduled for Q1 2027 and the 960PR planned for Q3 2027.

The company says both versions will arrive earlier than previously expected, while Huawei says performance improvements have exceeded the original roadmap.

"We're evolving our Ascend chip series on a one-generation-a-year cycle," said David Wang, deputy chairman and rotating chairman of Huawei.

"In 2028 and 2029, we will roll out the Ascend 970 and 980 chips, respectively. Thanks to the Tau (τ) Scaling Law, not only will their compute specifications continue to double, but you can also expect to see huge improvements..."

UnifiedBus connects 4,096 NPUs inside the Atlas 960E

UnifiedBus is Huawei’s interconnect architecture for linking processors, memory, and other components within large AI computing systems.

Huawei applies UnifiedBus in the Atlas 960E, a 960 variant that ties up to 4,096 NPUs and gives them access to shared memory across the SuperPoD.

It combines UnifiedBus with Hi-ONE optical technology, extending high-speed optical connections deeper into the computing system.

The approach uses near-packaged optics, or NPOs, placing optical components closer to processors to improve transmission speeds and reduce energy consumption.

Huawei says this design helps the SuperPoD move data between thousands of processors without relying entirely on conventional optical connections outside the computing system.

The Atlas 960E can use approximately 5,500 Hi-ONE units instead of about 48,000 conventional 800G optical modules.

This configuration can reduce power consumption by more than 550 kilowatts across the processor interconnection system.

The architecture also gives the Atlas 960E 8 EFLOPS of FP8 computing performance while Huawei says it supports models containing up to 10 trillion parameters.

Atlas 960E systems can also connect into SuperClusters containing 512,000 NPUs, with the architecture ultimately supporting million-processor configurations.

The larger systems are important because Huawei cannot rely solely on individual accelerator performance to narrow the gap with Nvidia.

Its strategy combines more processors with faster communication, allowing thousands of chips to operate within a shared computing environment.

UnifiedBus therefore becomes important because thousands of processors must exchange data continuously during demanding AI training and inference workloads.

• Scaleout's onboard AI let a BAE Systems ALMA drone rank, select, and strike an armored engineering vehicle in a January 2026 demo
• Humans set the mission, started the system, and kept a pilot in reserve, but did not ultimately use them during the test
• The test underscores how small AI models can essentially take over not only the target-selection part of the equation, but also perform the actual mission end-to-end without human intervention needed

A small drone flying over a snow-covered Swedish test range in January 2026 found several objects, ranked an armored engineering vehicle as the most valuable, and made a beeline for it before dropping an explosive.

There was no human involvement in the entire process, which included selection, targeting, and then neutralization by the drone, making it a considerably more autonomous approach than what the current battlefield uses.

The flight, part of BAE Systems Bofors' Affordable Loitering Modular Ammunition (ALMA) program with onboard AI supplied by Uppsala-based Scaleout Systems, resurfaced this month as AI safety became a key topic among industry and government stakeholders alike.

A growing threat and capability for modern militaries

Scaleout Systems tested the AI in question with a pilot on standby and a continuous connection throughout the exercise, during which the drone used AI to detect, identify, and geolocate potential threats.

Depending on how you frame it, you can argue that this is both a threat to modern militaries and a boon: an autonomous drone that can select targets without oversight or a backlink is essentially immune to jamming, offering a much higher likelihood of completing its mission than alternatives.

At the same time, it becomes riskier when it is completely cut off from communications, malfunctions, or fails to select the correct target without human oversight.

Interestingly, BAE Systems Bofors' own release on Winter Demo 2026 does not mention the strike at all. It frames the three-day Karlskoga event around roughly 20 startups and technology companies, with Swedish Defense Minister Pål Jonson joining a panel and saying, "We need more actors and entrepreneurs in the defense industry."

The test, despite lacking human interaction in its end-game, still relied on human input to some degree: a user pushed a button to start the exercise, and humans set the parameters overseeing the entire exercise.

This isn't the first machine to choose its own target, however; Israel's Harpy loitering munition, fielded since the 1990s, can autonomously detect and attack radar emitters whose locations were unknown at launch.

This is awkward timing for regulators, to say the least: on September 5, the 128 states party to the Convention on Certain Conventional Weapons agreed on a non-binding text on lethal autonomous weapons, after the United States and Russia pushed late changes, including to a section on human judgment and control.

As a result, Nicole van Rooijen of Stop Killer Robots told Reuters that three years of work had been "substantially diluted in the last hours." For now, Scaleout has shown that detection, ranking, and targeting can run on a small airframe without a data link, within parameters that people set.

Whether this makes it to a modern battlefield is increasingly likely, even as the Russia-Ukraine conflict becomes more reliant on autonomous machines. The more pressing question is one that will be answered sooner or later: how does such a drone make decisions if it is jammed from its source without any human oversight?

• Researchers report a new upgrade for lithium-ion
• Added graphitic carbon nitride can boost capacity
• It would mean higher power delivery for longer

With so many modern-day devices dependent on rechargeable lithium-ion batteries, scientists are hard at work trying to figure out ways of improving the tech — and a team from the Korea Research Institute of Chemical Technology (KRICT) has just announced a significant breakthrough.

As reported by The Korea Times, the KRICT researchers have succeeded in boosting the power delivery capacity of lithium-ion battery chemistry by adding a small amount of a material called graphitic carbon nitride to the cathode end of the battery.

When a battery is charging, the lithium ions get pushed out of the cathode, through an electrolyte material, to the anode at the other end. When the battery is in use, the ions flow back in the other direction.

One potential way of storing more energy in a battery of the same physical size and weight is to make the cathode thicker, but there's a problem: this also makes it harder for the lithium ions to move and deliver all of the energy held in the battery. The new innovation that was successfully tested here could solve that problem.

Capacity boosting

Microscopic imagery showing the cathode mix (Image credit: Eun et al., Exploration, 2026)

The researchers have reported some impressive numbers from their tests, though the tech is still at an early stage. Adding the graphitic carbon nitride to the cathode resulted in a 166% increase in capacity during high-rate discharge, and an increase in power density of up to 2.85 times.

Those figures don't mean a bigger capacity in mAh terms, but delivering more of a charge under heavy load, and for longer. In lithium-ion batteries, using a lot of energy quickly reduces how much total capacity you'll get from the battery compared to if you used the same amount of energy but more slowly.

In an EV, for example, this would mean that the vehicle could accelerate harder and sustain that high performance for longer without impacting total range as much. It would allow home batteries to better sustain their capacity if you use them with appliances that drain a lot of energy quickly, such as washing machines.

But smaller applications can benefit from it as well — drones can draw a lot of power suddenly to life themselves, or gaming and workstation laptops might suddenly draw a lot of power depending on the workload — they'd all benefit from tech that handles the power flow's effect on capacity better.

The tech could, in turn, result in batteries with thicker cathodes that do give you more mAh capacity in the same space, the researchers say — but that wasn't tested in this study. The next challenge is to try and scale up and expand the technology.

According to KRICT President Shin Seok-min, this alternative approach to lithium-ion batteries could potentially be used across electric vehicles, energy storage systems, and robotics — so it's very broadly applicable, especially for devices and systems that can take batteries that are physically larger.

• AWS deal with Nvidia adds 2 million Blackwell Ultra, Rubin, and Rubin Ultra GPUs, on top of more than 1 million committed in March 2026
• Totals more than 3 million Nvidia GPUs due before 2029, a ~1.9x multiple for each of Amazon's roughly 1.58 million employees
• Nvidia's CFO has said deliveries start this quarter, not in 2027, but neither company says what it counts as a GPU or what the deal costs

Amazon ended 2025 with about 1.58 million full- and part-time employees, most in fulfillment centers and delivery operations, with a much smaller share working in its AWS data center division (estimated at around 246,000).

By mid-2028, Amazon Web Services expects to have taken delivery of roughly twice that many new Nvidia GPUs, based on a recent deal it made with Nvidia.

The latest deal was announced on August 26, the same afternoon Nvidia reported quarterly earnings, and will see AWS plan to deploy two million additional Blackwell Ultra, Rubin, and Rubin Ultra GPUs across its global infrastructure in 2027 and 2028. This is on top of over a million Nvidia GPUs that AWS committed to at Nvidia GTC 2026 in March, and which Nvidia's Ian Buck told Reuters would ship from 2026 through 2027.

A mammoth deal that dwarf's Amazon's entire payroll

Together, this all amounts to over three million GPUs in play between the data center giant and the chip designer, which is also the world's most valuable company at the time of writing. At 3 million GPUs against a workforce of 1.58 million, that's nearly 1.9 Nvidia GPUs per employee.

This is only Amazon's commitment to Nvidia; it also has its own custom Tranium3 GPUs, which the company says are "nearly fully subscribed" or "almost fully booked" as of mid-2026.

At the same time, Amazon is not increasing job roles on its end; it cut around 30,000 corporate roles between October 2025 and January 2026, a restructuring its top HR executive, Beth Galetti, described as "reducing layers, increasing ownership, and removing bureaucracy."

Building on an already solid base

While AWS's current GPU base from Nvidia, consisting mainly of H100 and H200 GPUs, is smaller than Microsoft's 485,000 count, it is hardly a rounding error, with Omdia estimating that Amazon bought 196,000 Hopper GPUs in 2024 alone.

The interesting part is determining what counts as a GPU: Nvidia and Amazon's joint statement about 2 million GPUs across calendar years 2027 and 2028 might mean two very different things depending on how you count GPUs.

At GTC 2025, Nvidia said it would count Rubin-generation GPUs by compute die rather than by package, which is why the Vera Rubin rack was first unveiled as NVL144. It reversed course before CES 2026, and the flagship now ships as Vera Rubin NVL72, counting 72 two-die packages. To make matters a bit more complicated, Rubin Ultra, due in 2027, puts four compute dies in each package.

Whether the companies count each die as a separate GPU or the whole package as a GPU could make the numbers skew heavily in either direction. The deal also includes a national security element: approximately 100,000 GPUs are earmarked for US government AI workloads at Impact Level 6.

Amazon's limitation here is supply, not demand; AWS revenue rose 37% year over year to $42.2 billion in the second quarter, and its contracted backlog reached $496 billion, up from $364 billion three months earlier, as CloudTech News reported.

On Amazon's July earnings call, CEO Andy Jassy said capacity would fall short of demand this year and likely next, adding: "In fact, the demand we already have for 2028 is striking."

In addition to its own custom ASICs, including Tranium3, Amazon is gearing up for a few fiscal years of eye-watering AI infrastructure spend and AI demand combined, and it better be right: Amazon's free cash flow has already turned negative as it looks to fund its bid to be one of, if not the largest hyperscaler in play for the foreseeable future.

Via StorageNewsletter

My two sons are obsessed with 3D-printed toys, so a 3D printer is on their Christmas wish list, which makes this Amazon deal very appealing.

The Bambu Lab P1S is Amazon's number one best-selling 3D printer, and it just dropped to $499.99 (was $549.99). That's a $50 discount and the lowest price we've ever seen for the 3D printer.

• Shop more 3D printer deals at Amazon

The Bambu Lab P1S is a great introduction to 3D printing, with claims that it can be set up in just 15 minutes and offers hassle-free assembly for all skill levels. It supports up to 16 colors, so you can print endless, colorful creations, and it's fast too, with a max speed of 500 mm/s. It also includes a handy remote control and comes in an enclosed-body design with an activated carbon filter for improved performance.

Save on Amazon's best-selling 3D Printer

Lab P1S Combo 3D Printer: was $549 now $499.99
The Bambu Lab P1S is Amazon's number 1 best-selling 3D printer, and it's on sale for $499.99 - a record-low price. The Bambu Lab S1S takes just 15 minutes to set up, offers 16 colors for endless creations, and prints at lightning speed.View Deal

Keep in mind that today's discount on the Bambu Lab S1S is a limited-time offer and a record-low price. That $499.99 price tag is also an incredible deal for a high-quality 3D printer, as other models retail in the $600 to $800 price range.

• So-called signal sharing between Microsoft and Google uncovered the RedVDS marketplace was behind large-scale cyber fraud
• RedVDS sold access to virtual machines running unlicensed software from which cybercriminals launched scams and directed attacks
• Threat data was shared securely via the Global Signal Exchange (GSE), a non-profit organization

A major cybercrime marketplace has been taken out of action thanks to cooperation between Microsoft and Google via a secure threat data sharing platform.

Known as the Global Signal Exchange, the collaboration led to the take-down of the RedVDS, an online cybercrime mall supplying virtual machines for launching phishing, business email compromise attacks, and more.

The marketplace – which provided quickly-deleted and therefore virtually untraceable VMs – was meticulously monitored by Microsoft’s security team, Digital Crimes Unit, with both Microsoft and Google suspending the operation and action taken to seize domains and servers.

A secondary case featuring the GSE identified a Microsoft-impersonating tech support scam, resulting, highlighting the importance of secure thread data collaboration platforms.

Digital Crimes Unit

Finding massive scams is not easy. Even with its own Digital Crimes Unit, Microsoft has to rely on information sharing with third party organizations. In this case, the RedVDS marketplace was identified ahead of disruptive action taken by Microsoft in January 2026, which applied to courts in the UK and US to have the RedVDS web domains seized.

Data shared via the GSE enabled Google to follow suit, identifying related accounts on its networks and suspending them. Meanwhile, servers were impounded in Germany, and Europol took action against Europe-based RedVDS servers.

It’s an impressive account of cooperative, decisive action against cybercrime, a collaborative response to what has been described as a “$66 million fraud marketplace.”

An incredible 130,000 organizations were targeted by RedVDS-provided virtual machines between September and December 2025, with 191,000 Microsoft email accounts compromised during this period. All of this access to cybercrime was available from just $24 a month for a basic scam-ready virtual machine.

The importance of collaboration

The Global Signal Exchange is a UK-based non-profit, co-founded with Google in 2025 with the aim of providing real-time monitoring of the cybercrime supply chain.

"Fraud does not respect company boundaries, and no single organization ever sees the whole picture," noted Emily Taylor, CEO at Oxford Information Labs and Co-Founder of the Global Signal Exchange.

"That is exactly why we built GSE: to give trusted partners a secure way to share what they know, quickly. These two cases are a good example of GSE doing exactly what it was designed to do."

The Global Signal Exchange’s role in the sharing of information between Google and Microsoft has highlighted the importance in such collaborations. Both organizations have stated they plan to continue sharing information through the GSE, whose other partners include Meta, Amazon, Google, among many others.

We're gearing up for the second wave of major marathons this year: runners will be heading for the start line in Berlin at the end of the week (27 September), with the Chicago marathon following on 11 October and New York rounding out the season on 2 November.

Perhaps you're wondering what kind of footwear is propelling elite athletes to these kinds of times? I've hunted out the shoes worn by a selection of marathon winners, and rounded up the results below.

The Adidas Adios range crops up time and time again. Sabastian Sawe and Yomif Kejelcha both sprinted to stunning sub-2 hour success at the 2026 London marathon while wearing Adidas Adizero Adios Pro Evo 3 shoes (currently sold out in the US, but a restock is coming in the UK on 24 September.) Sawe wore Adidas Adios Pro Evo 2 to take the gold at the Berlin 2025 marathon, as did Benson Kipruto for his 2025 New York win.

Hot on Kipruto's heels in Berlin 2025 was Alexander Mutiso Munyao, wearing the Adidas Adizero Adios Pro 4. Incidentally, women's champ Rosemary Wanjiru was also wearing Adios Pro 4. In Chicago 2025, Hawi Feysa took the gold in the women's race wearing Adidas Adios Pro Evo 1.

Nike's winning line is the Alphafly. Jacob Kiplimo won the 2025 Chicago marathon in Nike Alphafly 3 — also the running shoe of choice of record-breaker Eliud Kipchoge.

It's not all about Adidas and Nike, though. John Korir has won two Boston marathons in a row while wearing prototype versions of the Asics MetaSpeed Sky, and Sharon Lokedi wore Under Armour Veliciti Elite 3 to defend her crown at Boston 2026. Hellen Obiri took the women's crown at the 2025 New York marathon in On LightSpray Cloudboom Strike — particularly notable for their laceless, 'spray-on' style.

If you're a keen runner and find yourself gripped by marathon fever, perhaps these are some models worth considering for your next long-distance race. But a couple of important caveats, before we start: First, if you're already in training for one of the upcoming marathons, you definitely don't want to be switching up your footwear in the run-up to the event. Second, these shoes are aimed at experienced long-distance runners; if you're not at that stage yet, head to our general best running shoe roundup for more beginner- and intermediate-friendly recommendations.

US: Marathon-winning running shoes

Jump down to UK shoes

Adidas Adizero Adios Pro 4 (men's)

Adidas Adizero Adios Pro 4 (women's)

Nike Alphafly 3 (men's)

Nike Alphafly 3 (women's)

Under Armour Velociti Elite 3 (unisex)

Adidas Adizero Adios Pro Evo 1 (unisex)

Asics Metaspeed Sky Tokyo (men's and women's)

Adidas Adios Pro Evo 2 (unisex)

On Lightspray Cloudboom Strike 2 White unisex)

UK: Marathon-winning running shoes

Adidas Adizero Adio Pro Evo 3 (restock soon)

adidas Adizero Adios Pro Evo 2 (unisex)

Nike Alphafly 3 Glam (men's)

Nike Alphafly 3 (women's)

Adidas Adizero Adios Pro 4 (men's)

Adidas Adizero Adios Pro 4 (women's)

adidas Adizero Adios Pro Evo 1 (unisex)

Asics Metaspeed Sky Tokyo (unisex)

UnderArmour Velociti Elite 3

On LightSpray Cloudboom Strike

• BAN says 87% of data center hardware isn't actually accounted for in most figures
• Each 1GW of capacity equals 70,000 tonnes of hardware
• AI hardware needs more frequent replacing

A new Basel Action Network (BAN) report has claimed the ongoing AI boom could create a new stream of e-waste, with retired equipment from AI alone likely contributing 395-617 million tonnes between 2025 and 2050.

To put that into context, BAN says that waste would fill 15-23 million 40ft shipping containers, which, placed end-to-end, would be enough to circle the earth six times over.

Ultimately, Moore's Law is to blame because developers are no longer able to achieve such great improvements in performance year-over-year, so they're having to introduce whole new types of technology.

AI will likely lead to massive amounts of e-waste

For example, AI companies are now having to deploy huge numbers of specialized AI accelerators alongside GPUs, cooling equipment, networking and more.

BAN also highlights that global data center Capex is expected to hit $7 trillion between 2025 and 2030, with $4.3 trillion of this attributed to electronic hardware.

The paper goes on to criticize earlier estimates for only including servers and accelerators, which it says is only around 13% of the picture. The other 87%, which spans networking, power distribution, storage/backup, and cooling, is largely unaccounted for.

And with each additional 1GW of data center capacity estimated to need around 70,000 tonnes of electronics, and 100GW of additional capacity anticipated globally by the end of this decade, the true scale starts to become clearer.

There's also refresh cycles, because BAN predicts that AI accelerators, servers and racks are replaced every 2.5 years, versus the usual five to seven years of a general-purpose server.

While these may be speculative estimations and not actual quantities, one thing's for certain – more data centers means more e-waste.

To minimise some of the impacts, BAN first suggests designing AI data center equipment to last longer in the first place. Recycling and second use is also a big topic, but the report doesn't go into whether AI-specific hardware can then go on to be used in lesser functionalities, such as being shoehorned into regular server facilities.

• Hackers took down a Flock camera and extracted a copy of its storage drive
• There, they were able to acess 1.6 million images and 27,000 videos taken over 21 days, despite claims that the encryption system should prevent this
• Morale is reportedly so low at Flock that many workers are ready to quit

Another day, another Flock controversy. As if things couldn’t get any worse for the embattled maker of automatic license plate recognition (ANPR) cameras, hackers have just managed to extract 1.6 million images and 27,000 video clips from one of its systems — something Flock has previously claimed should be impossible.

According to a report from 404 Media, members of a hacker collective named stegan0gram took down a Flock camera and managed to make a “near-complete copy” of its stored data.

When they did that, they found a treasure trove of privacy-compromising information and shared it with both 404 Media and the Distributed Denial of Secrets nonprofit, who then sent it on to Wired.

The report from 404 Media identified around 50,200 vehicles in Flock’s media files. The storage contained approximately 1.6 million images and more than 27,000 video clips taken over the course of 21 days. Older logs were either overwritten or inaccessible.

Perhaps more concerning, Flock’s cameras were able to detect people as well as vehicles and license plates, raising serious questions about how well the privacy of innocent members of the public is being protected. That said, 404 Media found no evidence of any kind of facial recognition being put to use.

The hackers said that they discovered two partitions when they accessed the Flock camera’s Android operating system. Inside one labeled “media,” they found a decryption key that unlocked a storage area housing the vast collection of images and video clips.

That all goes against public statements from Flock. The company has previously claimed that images and videos are only briefly stored on the device after being sent to Flock’s servers, and that no decryption keys were stored on the devices.

But with the hackers discovering stored media going back many weeks, plus the means to decrypt it, it appears that Flock’s statement might be inaccurate.

‘The writing is on the wall’

(Image credit: Getty Images)

As you can probably imagine given the near-constant barrage of negative press Flock has been facing in recent weeks, its own employees are reportedly starting to feel the strain. And with morale through the floor, many are seeking to leave the company.

Wired has reported that Flock recently opened up a voluntary separation program for employees that would offer a “generous” severance package to anyone who seeks a way out of the firm. The scheme is open until October 2, with decisions being announced on October 9.

And according to “people familiar with the program but not authorized to discuss it publicly” cited by Wired, morale is so low at Flock that “a significant number of the startup’s roughly 1,500 employees may try to depart.”

That backs up comments from Flock’s own CEO, who said on the All-In podcast in August that “internal moral” has been the biggest casualty amid escalating hostility to the company’s devices.

For many Flock workers, the severance offer is likely to come at a good time. As one apparently disgruntled employee told Wired, “When you see enough of the vandalism [aimed at Flock cameras], you know the writing is on the wall.”

If you're the kind of person who only dabbles in ChatGPT now and again, you might be getting a bit bored.

Maybe you use it to proofread work, plan a trip or make images sometimes. But with the cost of a subscription, and growing concerns about AI’s environmental impact, its effect on our cognitive abilities and what it might mean for jobs, you might reasonably be wondering: is ChatGPT actually useful enough to justify using it?

Well, you're not alone. Plenty of AI users have been telling me the same. So I went looking through Reddit threads where ChatGPT users were sharing the more unusual and genuinely useful ways they use it. And I found a few that might actually be worth experimenting with.

One Redditor asked: What's the most unexpectedly useful thing ChatGPT has done for you? Another posted: What are the most useful ChatGPT features and use cases that most people do not know about?

So, I went through the most upvoted replies looking for ideas that went beyond the usual writing, brainstorming and summarizing suggestions. Some of them made me think about ChatGPT's capabilities in a different way.

But a quick caveat before we start, I haven't tested every suggestion below and we know ChatGPT can make mistakes, so be cautious when an answer could affect your health, safety and finances. Think of these more as inspiration for using ChatGPT more creatively rather than fully vetted recommendations.

1. Declutter your junk drawer

(Image credit: Future)

I know this one might seem a bit silly. And maybe you thought you'd have to wait until ChatGPT gets a robotic body to help you tidy up. But I promise it might make your dull organization tasks a little less boring.

One Redditor said they use ChatGPT to: “organize and declutter a junk drawer or a box of power adaptors just using pictures.”

Rather than sitting there examining every mysterious cable and adapter you’ve accumulated over the past decade, you can just take a photo and ask ChatGPT to help you work out what everything is and, most importantly, how to sort it.

You can ask it to suggest categories for what you’ve decided to keep and where each group should live. The same trick could work for a toolbox, kitchen cupboard, art supplies or that box of cables you’re keeping because you're convinced one of them will become extremely important as soon as you throw it away.

So, I decided to give it a go. I took a picture of a box that sits on my desk packed full of contact lenses, medication, a bunch of different cables, notes and hair ties.

The advice was helpful, ChatGPT told me what each thing was and how to sort everything. The most useful thing it then did was send my photo back to me with annotated and color-coded notes so I could quickly go about sorting. This was perfect for my ADHD brain.

(Image credit: Future)

There are obvious limits here, you wouldn't throw away something valuable just because ChatGPT told you too. But as a starting point for tackling a chaotic collection of stuff, it might be helpful.

2. Talk yourself out of buying something

AI shopping tools are usually designed to help you find more things to buy. But one Redditor uses ChatGPT for the opposite purpose: "[It] talks me out of purchases. I upload a potential purchase and ask it to talk me out of it. I’ve saved quite a bit of money!"

I love the reversal here. Instead of asking ChatGPT whether you should buy the $200/£200 headphones currently sitting in your basket, tell it you want them and ask it to make the strongest possible case for why you shouldn't buy them.

You could give it the price, tell it what you already own, explain what you think the new product will improve and ask questions, like: What problem am I actually solving here? Do I already own something that does this? What are the strongest reasons to wait a month?

I gave this one a go. I love working out outside and at the gym, but in the depths of winter I find it really hard to stay motivated and I'm considering getting a small treadmill or walking pad. There's nothing wrong with them, but it's not my favorite way to workout, I don't have much room for one and the good ones aren't cheap. I really need someone to talk me out of getting one. Would ChatGPT be up for the challenge?

Here's how it responded:

"Don’t buy it. You already know the case against it. You don’t enjoy treadmill workouts, don’t have much space, and don’t want to spend the money. You’re considering buying a fairly large object to solve a problem that exists for a few dark months a year."

The response felt a little short and sickly sweet. If my finger was genuinely hovering over the "buy it now" button I'm not sure this would have stopped me. But I do love the idea of using AI to actually add some perspective and friction back into our lives.

3. Streamline your skincare routine

(Image credit: Future)

This one is similar to the idea above, but given how much people spend on skincare these days and how confusing the advice is about how to build the perfect routine, I think it's well worth testing out.

One Redditor said: “I gave it all my skincare. It cross referenced, compared and analyzed all the ingredients and items I had and recommended what I don’t need based on the overlaps. Saved me a lot time and money.”

In other words, don't ask ChatGPT what else you should buy. Show it what you already own and ask where you're duplicating things and if something does genuinely need adding.

For skincare, that could mean listing your cleansers, serums and moisturizers and asking it to identify products containing similar active ingredients. You could potentially apply the same principle to all sorts of collections too.

I took a quick snap of my core skincare staples and asked it to analyze the ingredients, tell me if there's overlap and ask what I could add, but stressed I'm on a budget and want to keep things simple.

It responded with a detailed breakdown of each product, its ingredients and what they do. It was also very measured about suggesting new products, and even told me my eye cream wasn't necessary. I was also happy it included sources for every claim, which it doesn't always do but that's important for skincare.

(Image credit: Future)

I then asked it to present this information on the image itself because I found that really useful for the junk drawer sorting example, and it was such a handy visual.

There are limitations here too. ChatGPT isn't a dermatologist, and ingredients alone don't determine how suitable a skincare product is for you. But I like the broader idea of using AI to audit your consumption rather than constantly optimize it by adding more.

4. Identify plants, animals and trees

(Image credit: OpenAI / Apple)

“I use it a lot to ID plants and animals,” one Redditor explained, adding that although ChatGPT occasionally produces “a weird incorrect response”, it's usually quite good.

This is another use for ChatGPT's camera capabilities that could be easy to overlook if you mostly interact with it through text.

Photograph a plant, insect or bird and ask ChatGPT what it thinks you're looking at. Better still, ask it what features it's using to make the identification and what similar species it could be confusing it with.

I tried this out with a bunch of trees, asking ChatGPT to identify them based on their leaves. I then fact-checked everything and it identified them all correctly. It also provided details about the history of the trees too. Most of the trees in my area are pretty standard, like Oak, Ash and Beech, so it'd be interesting to see how it fares with trees, plants or insects that are a little more unexpected.

If it’s bird song you’re interested in though, I highly recommend using the identification app Merlin, it’s one of my most-used and much-loved apps.

And also remember that ChatGPT doesn’t always get image identification right, and you shouldn't eat, touch or otherwise interact with a potentially dangerous plant, fungus or animal based purely on a chatbot's identification.

5. Plan a walking tour

(Image credit: Future)

This tip came up in several forms. One Redditor said: “If I'm taking a walk in an interesting place, like a historic district, I show it a particular building or place and it gives me the history behind it. Like having a walking tour guide.”

Another user takes a screenshot of a hiking route and asks ChatGPT for interesting facts about the geology, history and place names they'll encounter along the way.

This is one of those ideas that feels obvious once somebody else has suggested it. Before a walk, you could upload a map and ask for five interesting things to look out for. Or when you encounter an unusual building, monument or landscape feature, take a photo and ask what you're looking at.

I’d be wary of confidently repeating every historical fact ChatGPT gives you to your hiking companions without checking it first. We know that chatbots can still invent plausible-sounding details. And never rely on it for route planning, especially though difficult terrain. There have been far too many horror stories about AI sending people in completely the wrong direction recently.

(Image credit: Future)

I tried this one myself and shared a Google Maps screenshot of Saltaire, a picturesque village in the UK. I asked ChatGPT to mark up the map with the best things to see on a trip there.

It showed me all of the key things to visit in the area. Granted there wasn't anything on here I couldn't have found from a very quick search. But it's still handy if you're in a new area for a few hours and want a quick overlay of the map you already have.

6. Troubleshoot a broken appliance using photos

(Image credit: OpenAI / Apple)

Plenty of people already ask ChatGPT technical questions. What I hadn't really considered was combining those questions with its ability to “see” what you're seeing.

One Redditor described doing exactly that when their dishwasher stopped working. They wrote: “Fixing appliances. Our dishwasher stopped working. I gave it the error message and took a photo of the dishwasher and it walked me through every step to fix it.”

They say ChatGPT eventually helped them identify a blockage and suggested ways to clear it. This seems useful for those frustrating household problems where you don't know the correct name for the thing you're looking at. Instead of trying to Google “the little plastic thing underneath dishwasher filter is broken”, you can simply show ChatGPT the little plastic thing.

Give it the appliance's make and model if you can, photograph the problem and provide the exact error message.

I tried this with the water heating system in my new flat because my landlord gave me the wrong manual. ChatGPT needed a lot of additional details but did end up giving me the right instructions for basic tasks, like setting a timer.

There’s a big safety caveat here. I wouldn't follow AI-generated instructions involving electricity, gas, dangerous machinery or anything else where getting it wrong could injure you. But for basic troubleshooting, identifying components and understanding error codes, the combination of text and images could be genuinely useful.

7. Find the glasses that'll suit your face

One Redditor uploaded a photograph of themselves and asked ChatGPT for help choosing prescription glasses:

“It helped me identify prescription eyeglass frames to fit my face. Uploaded a picture and it gave me the size, shape and website with model and frame number. Turns out I was wearing the wrong shape most of my life.”

I need new glasses so decided to try this idea out for myself.

(Image credit: Future)

I'd take the idea of a "right" or "wrong" glasses shape with a big pinch of salt. Style rules about which frames supposedly suit particular face shapes are subjective.

But doing this did help me start thinking about which glasses I should try on when I have my appointment at the optician's in a few weeks and narrow down the enormous number of options that are available.

You could also show it frames you already like and ask it to describe their characteristics so you know what terms to search for.

I think what's really interesting about all of these suggestions is that most of them aren't about building the cleverest or most elaborate prompt. Instead, they started with something in everyday life that needed identifying, organizing, checking, fixing or remembering.

So if you want to get creative about how ChatGPT could be more useful, maybe you need to look around more and ask, what do I really need help with?

• ShinyHunters hacked rival gang Cl0p, stealing source code, logs, and onion service keys
• The gang defaced Cl0p’s site via Grav CMS flaw, posting their logo and taunts about past threats
• Feud recalls Conti’s collapse in 2022, raising risk of escalating “cyber war” between criminal groups

Infamous cybercriminal gang Cl0p has seemingly been hacked by an even more infamous data leak collective, ShinyHunters.

The attack is still being pieced together, but it would seem we have true hacker beef on our hands - which just might escalate into a full-blown cyber war.

ShinyHunters has added Cl0p to their data leak site, giving the hacking group 72 hours to pay a ransom or see all their files leaked into the cybercriminal underbelly. The files allegedly stolen in the attack include source code, Grav CMS plugins, system logs, and other information, Cybernews reports. We don’t know how much money ShinyHunters are asking to keep Cl0p’s files private.

Still downloading

"The data we stole includes source codes, gravCMS plugins, and other things. We are still downloading and reviewing them," ShinyHunters allegedly told BleepingComputer.

The group also said they stole everything in the server’s /var/log directory, including system activity records, authentication logs, and IP addresses associated with connections to the server. We doubt this could lead to the identification of any Cl0p members, and even if it could, it would mean very little since the members are likely Russian and thus mostly free to conduct their operations.

Doxxing might help defenders disrupt the group’s infrastructure, though.

Whether or not that will be needed, given ShinyHunters’ activity, remains to be seen. The attackers also said they stole the private keys for Cl0p’s Tor onion service:

"We have their onion keys. So if they kick us out it wouldn't matter at all because we control the private keys to host the same exact onion URL," the group said.

On top of it all, ShinyHunters defaced Cl0p’s website. Allegedly, the group found an unauthenticated file upload flaw in the Grav CMS Cl0p was using, gaining access to their target’s servers, website, and more. Cybernews confirmed the website had been defaced and now holds an ASCII image of ShinyHunters’ logo, and a link to their Tor site. It also displays a large message:

"THIS SITE HAS BEEN PWN3D BY SHINYHUNTERES #Skids10p - Maybe don't try to threaten us next time.” Furthermore, it says “rooting your systems since '19 ;)".

The “maybe don’t try to threaten us next time” message seems to refer to a threat that a Cl0p member made earlier. Apparently, this person threatened violence against ShinyHunters, as well as to expose their identities to the public. The remarks were made during Cl0p’s 2025 Oracle E-Business Suite attacks.

"During the Oracle EBS campaign they ran and stole from me last year, someone from cl0p personally messaged me and said, and I quote (translated from Russian): I have more money than you and all of your people combined, I'll kill you soon," ShinyHunters told BleepingComputer.

Dog eat dog

Cybercriminals often work together. They share resources, rent each other’s services, and often complement their campaigns. However, that doesn’t make them “friends” or “colleagues”, in any way. It is usually just a marriage of convenience that falls apart at the slightest sound of trouble. The last big “feud” we’ve seen was back in 2022, at the very start of the Russian invasion of Ukraine. Back then, a group called Conti publicly declared its “full support” for the Russian government and threatened to use its resources against anyone conducting cyberattacks on Russian infrastructure.

The announcement was a fiasco - many of Conti’s affiliates stopped working with the group, and others were openly attacking it. Soon after, a Ukrainian researcher and an alleged affiliate leaked more than 60,000 messages, exposing the group’s operations and internal organization.

All of this eventually led to Conti’s collapse later that year. However, its members did not disappear - they splintered into multiple groups such as Black Basta, Royal, and Quantum. Some of the crooks engaged in Conti are allegedly still active today.