Feed aggregator

Hints and answers for Connections for March 25, #653.

Here are hints and the answer for today's Wordle No. 1,375 for March 25.

Here are hints and the answers for the NYT Connections: Sports Edition puzzle, No. 183, for March 25.

I was talking to an old friend about AI – as one often does whenever engaging in causal conversation with anyone these days – and he was describing how he'd been using AI to help him analyze insurance documents. Basically, he was feeding almost a dozen documents into the system to summarize or maybe a pair of lengthy policies to compare changes. This was work that could take him hours, but in the hands of AI (perhaps ChatGPT or Gemini, though he didn't specify), just minutes.

What fascinated me is that my friend has no illusions about generative AI's accuracy. He fully expected one out of 10 facts to be inaccurate or perhaps hallucinated and made it clear that his very human hands are still part of the quality-control process. For now.

The next thing he said surprised me – not because it isn't true, but because he acknowledged it. Eventually, AI won't hallucinate, it won't make a mistake. That's the trajectory and we should prepare for it.

The future is perfect

I agreed with him because this has long been my thinking. The speed of development essentially guarantees it.

While I grew up with Moore's Law, which posits a doubling of transistor capacity on a microchip roughly every two years, AI's Law is, putting it roughly, a doubling of intelligence every three-to-six months. That pace is why everyone is so convinced we'll achieve Artificial General Intelligence (AGI or human-like intelligence) sooner than originally thought.

I believe that, too, but I want to circle back to hallucinations because even as consumers and non-techies like my friend embrace AI for everyday work, hallucinations remain a very real part of the AI, Large Language Model (LLM) corpus.

In a recent anecdotal test of multiple AI chatbots, I was chagrinned to find that most of them could not accurately recount my work history, even though it is spelled out in exquisite detail on Linkedin and Wikipedia.

ChatGPT had me working at a place I've never worked (left). DeepSeek couldn't get the dates right (center), and ClaudeAI (right) also had timeline issues. (Image credit: Future)

These were minor errors and not of any real importance because who cares about my background except me? Still, ChatGPT's 03-mini model, which uses deeper reasoning and can therefore take longer to formulate an answer, said I worked at TechRepublic. That's close to "TechRadar," but no cigar.

DeepSeek, the Chinese AI chatbot wunderkund, had me working at Mashable years after I left. It also confused my PCMag history.

Google Gemini smartly kept the details scant, but it got all of them right. ChatGPT's 4o model took a similar pared-down approach and achieved 100% accuracy.

Claude AI lost the thread of my timeline and still had me working at Mashable. It warns that its data is out of date, but I did not think it was 8 years out of date.

What percentage of AI answers do you think are hallucinations?March 24, 2025

I ran some polls on social media about the level of hallucination most people expect to see on today's AI platforms. On Threads, 25% think AI hallucinates 25% of the time. On X, 40% think it's 30% of the time.

However, I also received comments reminding me that accuracy depends on the quality of the prompt and topic areas. Information that doesn't have much of an online footprint is sure to lead to hallucinations, one person warned me.

However, research is showing that models are not only getting larger, they're getting smarter, too. A year ago, one study found ChatGPT hallucinating 40% of the time in some tests.

According to the Hughes Hallucination Evaluation Model (HHEM) leaderboard, some of the leading models' hallucinations are down to under 2%. Older models like Meta Llama 3.2 are where you can head back into double-digit hallucination rates.

Cleaning up the mess

What this shows us, though, is that these models are quickly heading in the direction my friend predicts and that at some point in the not-too-distant future, they will have large enough models with real-time training data that put the hallucination rate well below 1%.

My concern is that in the meantime, people without technical expertise or even an understanding of how to compose a useful prompt are relying on large language models for real work.

Hallucination-driven errors are likely creeping into all sectors of home life and industry and infecting our systems with misinformation. They may not be big errors, but they will accumulate. I don't have a solution for this, but it's worth thinking about and maybe even worrying about a little bit.

Perhaps, future LLMs will also include error sweeping, where you send them out into the web and through your files and have them cull all the AI-hallucination-generated mistakes.

After all, why should we have to clean up AI's messes?

You might also like

• The ultimate AI search face-off - I pitted Claude's new ...
• Researchers want to give some common sense to AI to turn ...
• The surprising reason ChatGPT and other AI tools make ...

Tulsa has great access to high internet speeds and top internet providers. Here's which ISPs we recommend.

A new study shows that the quality of a person's microphone in a video meeting affects how the speaker is perceived by others.

(Image credit: Smith Collection/Gado)

Apple Sport Band review

The Apple Sport Band is a basic, albeit high-quality Apple Watch strap that’s ready for all activity, whether that be a stroll in the park or a run around the block. At $49 / £49, it’s not the cheapest Apple Watch band in the world, but its price is very much in-line with other official first-party accessories, like the Sport Loop, for instance.

This band is made of fluoroelastomer – ideal for sports due to its inherently sweat and waterproof nature. When I went on walks and did a home workout, I never found that the band was irritating on my skin or accrued any kind of damage.

On top of that, it’s super comfortable for everyday use. Even fastening the band on is quick and easy – just poke the aluminum pin through one of the strap’s holes and you’re good to go. That combination of comfort and convenience is a must-have for all of the best Apple Watch bands.

Something I would have liked to see, though, is a few more holes in the strap for fitness activity. That’s something that I loved about the Apple Nike Sport Band, for example, because it offers added breathability when you’re working up a sweat. Another aspect of the Nike Sport Band I appreciated was that it is composed of 30% recycled fluoroelastomer, but the Apple Sport Band doesn’t come with this sustainability pledge.

But enough moaning – I want to get back to the positives again. This band is available in a wide array of colorways – seven to be precise. I tested out the Plum variant, but this strap also comes in: Black; Denim; Lake Green; Light Blush; Starlight; and Stone Gray. All of these look fairly understated, making for a more discreet look, if that’s what you’re after.

In addition, this strap is compatible with three different case sizes: 40, 42 and 46mm. You can grab it either in small-to-medium (130-180mm) or medium-to-large (150-200mm).

Even if this band comes in at $49 / £49, you’re still getting value for money. Sure, you could opt for a third-party alternative – like the Yoohoo Adjustable Metal Strap – for less than $15 / £15, but this will be more prone to scuffs and won’t be so easy on the eye. Meanwhile, the Apple Sport Band has a satisfying material, reliable fastening method and classy looks. That makes it well worth the extra spend as far as I’m concerned.

All in all, I think the identically priced Apple Nike Sport Band beats this one out thanks to its more airy design and vibrant looks. But for anyone that likes a more reserved aesthetic, I’d say the Apple Sport Band is a great option for you. It has that real quality feel and it's super simple to slip on if you’re on-the-go. That alone earns it a recommendation in my book.

(Image credit: Future) Apple Sport Band review: Price & specs

(Image credit: Future) Should I buy the Apple Sport Band? Buy it if…

You want an understated look
The Apple Sport Band comes in seven color options. None of these are dazzling and vibrant, but all of them play into a smooth, discreet look very nicely. When testing out the plum band, for example, I never felt that it stuck out like a sore thumb against my clothing.

You want to get fastened up quickly and easily
Just poke the aluminum pin through one of the Apple Sport Band’s holes and bam! It will lock into place straight away. There’s no frustrating buckle or unorthodox contraption to wrestle with here, just pop and go. Simple, yet effective.

Don't buy it if…

You want a cheap replacement strap
If you’re on the hunt for the ultimate bargain option, this ain’t it, I’m afraid. The Apple Sport Band will set you back $49 / £49 – by no means cheap for an Apple Watch accessory. Sure, that’s in line with the Nike Sport Loop, for example, but there are some really cheap third-party alternatives, like the CeMiKa Silicone Sport Straps if you’re looking for savings.

You want a super-classy band
If you want a super swanky band, then you’re in luck – there are plenty of premium alternatives to be discovered on the Apple Store. Options like the Milanese Loop or Hermès En Mer, for instance, carry that air of high class that some of you may be craving – but these options will set you back by at least $100 / £100.

Apple Sport Band review: Also consider

Apple Nike Sport Band
This one’s extremely similar – it even comes in at the same price. But if you want a bit more flair in your watch band’s design, the Apple Nike Sport Band is the perfect option. It has more vibrant, loud color options, eye-catching fluoroelastomer flakes throughout and breathable holes when you need to sweat it out. It’s been my go-to band since I started using the Apple Watch SE, so why not give it a try? Read our full Apple Nike Sport Band review.

• HorizonOS v76 has rolled out to the public test channel
• This beta release is teasing new fetaures Meta is working on
• In future updates we can expect major UI changes, and shareable windows

Meta has just launched the HorizonOS v76 update to its public test channel, and the beta software is already teasing some massive changes for how you can use your Meta Quest 3 headset to virtually socialise.

Firstly, Meta is putting your Horizon Avatera front and center in video calls, finally unlocking the selfie camera – a feature it first teased back at Meta Connect 2022. You could previously take Zoom meetings from your virtual workspace, but with update v76, you’ll be able to use your Meta avatar in more casual video calls through WhatsApp and Messenger.

Avatar Selfie Cam UI in Meta Quest/Horizon OS v76 PTC.Doesn't seem to be enabled out of the gate though. pic.twitter.com/zOG0aya5NiMarch 22, 2025

In Settings, you can see your Selfie cam options to adjust how narrow or wide the virtual camera is, and you can select a static background that will appear behind your character.

Then, when you join video calls while using your headset, other people will see your avatar moving as you move. However, people who have tested the in-development tool say it is still limited.

‘In-development’ is definitely the key description here, as Selfie cam still feels very limited so it might take a little while before it reaches the wider HorizonOS public release.

Further, when it does, Meta might move it to be an ‘experimental feature,’ which is a designation given to features that are available in the full HorizonOS release, but that might be a little buggy still.

Strings in Quest/Horizon OS v76 PTC suggest that Meta is working on the ability to share windows with other users in Horizon Home (and possibly Worlds).This will likely work similarly to SharePlay on visionOS. pic.twitter.com/ZudymM05XJMarch 22, 2025

Update v76 in the PTC also hides details about the ability to share your screen with other Meta Quest users.

The feature isn’t live yet, but code strings (discovered by Luna) suggest that 2D window panels will gain a ‘share’ and ‘unshare’ button so you can show other people in Horizon Home or Horizon Worlds (and maybe other multiplayer apps) what you’re looking at in your browser.

The Quest 3 already has the ability to screenshare YouTube content, and this release seems like a more general rollout of that bespoke feature so other 2D apps can be shared.

Given its current state in the PTC update, screen sharing might be an update or two away. However, when it does arrive, it might be joined by a massive UI overhaul.

Codenamed ‘Navigator’ Luna shared a short five second long clip of a tutorial for the new layout – which Meta demo’d at Meta Connect 2024.

Meta teased "the future of Horizon OS" at Connect today, showing a concept of a complete redesign.Details here: https://t.co/nYX2CfKeXt pic.twitter.com/Knavsn3p54September 26, 2024

Luna added that it’s expected to drop in v77 or later, so it’s still a release or two from launch, but these first hints suggest this overhaul’s launch is approaching.

We’ll have to wait and see if this UI overhaul is what Quest 3 has been needing all along or one of those terrible changes that'll have us begging Meta to put everything back the way it was.

From what we’ve seen, it should be the former, but we won’t know until the Navigator UI is available for everyone to test (hopefully later this year).

You might also like

• I can’t believe it either, Hitman on PSVR 2 is actually, finally a great VR port of the World of Assassination trilogy
• ‘Never say never’ – we ask Hitman developer if Project 007 is coming to VR
• Latest Snap Spectacles update teases an exciting AR future that I can't wait for

• If you have a Hatch Restore sunrise clock and a Hatch+ subscription, you can now start your day with The Office theme song
• Hatch and Peacock collaborated on a new sound and two light effects
• You can get ready for bed with the Dunder Mifflin White Paper Light

We’ve seen all sorts of collaborations celebrating The Office (US) – aka one of the most timeless and iconic sitcoms to ever broadcast or stream – but this one might be the most out of left field, in the best way possible.

Sure, we've seen collaborations before: Mattel dropped an incredible Polly Pocket that lets you carry the TV show's office set, including the conference room, in a mug with countless characters inside, and Lego will let you build the actual office from The Office. But now, you can both wake up to the classic theme song, and even be greeted with light that emulates a Dunder Mifflin sheet of paper.

How? Well, all you’ll need is a Hatch Restore sunrise alarm clock and you’re then ready for Bears, Beets, and Battlestar Galactica at any time. Thanks to a new partnership between Hatch and Peacock – one of the best streaming services around and home to every episode of The Office in the United States – you can enjoy the TV show’s theme song as an alarm sound or bedtime tune, as well as two light effects.

(Image credit: NBC)

Like any other sunrise alarm clock, the Hatch Restore – now in its third generation – aims to help you get to bed easier and wake up more smoothly with scenes. The latter combines audio – either a cozy tune to help encourage a window or a peaceful set of tones to wake up – and lighting, mimicking a sunrise or sunset.

This collaboration with The Office goes beyond the expected, though – waking up or settling down to the jungle is pretty fun and about where we would have expected this collaboration to start and finish.

However, the design team at Hatch also created lighting effects to help you start or end your day. In the morning, the new wakeup cue plays the theme song along with a “Scranton Commute Sunrise Light,” and for bedtime, you’ll hear the jingle as well as a “Dunder Mifflin White Paper Light.”

Whether or not you are superstitious – or if you’re just a little stitious – you won’t want to wait to give these The Office themed songs and lighting effects ago, as they are only here for a limited time. Available via the Hatch+ subscription in the United States or Canada, you can access The Office themed content, which will only be available for a month.

Hatch+ is available at $4.99 a month or $49.99 a year and gives you access to a ton of sleep content, including podcasts, soundscapes, and various lights.

(Image credit: Hatch)

While it’s only here for a limited time, this Hatch x The Office collab might feel like Pretzel Day for some, and who knows, waking up with the theme song and a Scranton sunrise might just let you start your day off a little better.

The Hatch Restore 3 – the latest generation sunrise alarm clock from the brand – is up for order at $169.99 in Putty, Greige, or Coacoa, and you get a free trial for Hatch+. Considering the previous generation has been discounted for previous Amazon shopping holidays, we’d keep an eye on it and see if it drops to a better price for Amazon’s Big Spring Sale.

You might also like

• Forget Netflix, I tuned into Peacock to watch the SNL 50 special and it went off without a hitch – here’s why it’s built for live-streaming
• I can't wait to see my favorite The Office character returning for Peacock's new spin-off series
• The best wake-up lights 2025: sunrise alarms tested by experts

• Nvidia’s DGX Station is powered by the GB300 Grace Blackwell Ultra
• OEMs are making their own versions – Dell’s is the Pro Max with GB300
• HP’s upcoming GB300 workstation will be the ZGX Fury AI Station G1n

Nvidia has unveiled two DGX personal AI supercomputers powered by its Grace Blackwell platform.

The first of these is DGX Spark (previously called Project Digits), a compact AI supercomputer that runs on Nvidia’s GB10 Grace Blackwell Superchip.

The second is DGX Station, a supercomputer-class workstation that resembles a traditional tower and is built with the Nvidia GB300 Grace Blackwell Ultra Desktop Superchip.

Dell and HP reveal their versions

The GB300 features the latest-generation Tensor Cores and FP4 precision, and the DGX Station includes 784GB of coherent memory space for large-scale training and inferencing workloads, connected to a Grace CPU via NVLink-C2C.

The DGX Station also features the ConnectX-8 SuperNIC, designed to supercharge hyperscale AI computing workloads.

Nvidia’s OEM partners - Asus, HP, and Dell - are producing DGX Spark rivals powered by the same GB10 Superchip. HP and Dell are also preparing competitors to the DGX Station using the GB300.

Dell has shared new details about its upcoming AI workstation, the Pro Max with GB300 (its DGX Spark version is called Pro Max with GB10).

The specs for its supercomputer-class workstation include 784GB of unified memory, up to 288GB of HBM3e GPU memory, and 496GB of LPDDR5X memory for the CPU.

The system delivers up to 20,000 TOPS of FP4 compute performance, making it well suited for training and inferencing LLMs with hundreds of billions of parameters.

HP’s version of the DGX Station is called the ZGX Fury AI Station G1n. Z by HP is now one of the company’s product lines, and the “n” at the end of the name signifies that it’s powered by an Nvidia processor - in this case, the GB300.

HP says the ZGX Fury AI Station G1n “provides everything needed for AI teams to build, optimize, and scale models while maintaining security and flexibility,” noting that it will integrate into HP’s broader AI Station ecosystem, alongside the previously announced ZGX Nano AI Station G1n (its DGX Spark alternative).

HP is also expanding its AI software tools and support offerings, providing resources designed to streamline workflow productivity and enhance local model development.

Pricing for the DGX Station and the Dell and HP workstations isn’t known yet, but they obviously aren’t going to be cheap. Pricing for the tiny DGX Spark starts at $3,999, and the larger machines will cost significantly more.

You might also like

• These are the best mobile workstations you can buy right now
• And these are the best business laptops on the market
• Nvidia’s DGX Station is a powerful supercomputer-class workstation PC

Thomas Tuchel's Three Lions look to claim another win at Wembley.

Democrats need to flip three seats to take back the House next year – and the path to a majority likely runs through districts President Trump carried. Democrats who won alongside Trump offer their prescription for a party they say needs to make big changes.

(Image credit: CHARLY TRIBALLEAU/AFP via Getty Images)

The Volunteer State has plenty of fast and reliable internet options. From Nashville to Knoxville, here are CNET's expert picks for the best internet service providers across Tennessee.

Is it ever safe to flush a "flushable" wipe? We asked a plumber to find out.

There are several providers in South Carolina that offer fast speeds, reliable connections and affordable plans. Here are our top picks.

On paper, it's a contender. On screen, it is not.

• Russia's Roskomnadzor has intensified internet blocks across the country
• Google has reportedly received at least 47 removal request orders so far
• Cloudflare subnets were also temorarily blocked, causing large-scale disruptions of major websites

Internet users in Russia have been suffering a new wave of online disruptions lately, with VPN apps and a major DNS server provider being the targets.

Starting March 12, 2025, Russia's censor body Roskomnadzor has reportedly hit Google with at least 47 removal orders of some of the best VPN apps from its Google Play store.

A few days later, on March 20, Cloudflare subnets were also temporarily blocked. This causes large-scale outages across multiple Russian regions, with many popular websites going dark. Some VPN services also experienced connectivity issues due to the DNS blocking.

Digital rights experts across the country now fear that's just the beginning of further blocks.

An ever-more restricted internet

Android and Windows users in Russia could soon be prevented from downloading almost 50 of the most popular VPN apps in an escalation of VPN censorship in the country. (Image credit: Photo Illustration by Idrees Abbas/SOPA Images/LightRocket via Getty Images)

Roskomnadzor's fight against VPN apps is nothing new but the scale of Russia's VPN blocking efforts keeps intensifying.

So, while almost 200 VPNs are already blocked across the country, Roskomnadzor has presented Google with some new removal orders.

A Russian journalist has counted 47 new requests since March 12, with the likes of HideMyNetVPN, Proxy Shield VPN, and Secure VPN being among the targets. It isn't yet clear if Google has already complied with such orders at the time of writing.

This new wave of VPN app removal from the Google Play Store follows the same modus operandi that has brought Apple to kill around 60 VPN apps from its App Store in Russia between July and September, bringing the total to 98 unavailable applications in the Big Tech giant's official store.

The orders appear to have been issued under a law enforced in March last year, which criminalizes the spread of information about ways to circumvent internet restrictions – VPNs included.

Website outages were observed across Russia this week, with regulators attributing them to issues with foreign servers. Observers said the problems might be tied to Russian government moves to block Cloudflare services https://t.co/bd0gu5g2c1March 20, 2025

Russia's latest crusade against VPN services came at the same time Cloudflare subnets (a group of over 500k IP addresses within the network) were also blocked across the eastern part of the country, from the Urals to Primorye.

Around 1.5 million IP addresses were impacted – a technical expert for Russian digital rights advocacy group Roskomsvoboda told TechRadar. The likes of TikTok, Steam, Twitch, Epic Games, DeepSeek, Duolingo, and mobile operator sites were all inaccessible without a VPN.

"VPN services have also experienced problems, as they often have their own management infrastructure linked to Cloudflare," Roskomsvoboda told TechRadar, noting that Warp VPN, which is developed by Cloudflare, also stopped working.

Commenting to a Russian news agency, Roskomnadzor said officials "will conduct scheduled technical checks of the use of foreign server infrastructure by Russian services and telecom operators."

While the incident has now been resolved, Roskomsvoboda experts warn that the long-term plan may be blocking Cloudflare completely as Iran did.

"Judging by the scale of these ‘exercises’, this may happen quite soon," the expert told Techradar, explaining that all the mobile apps and online services relying on Cloudflare as a Content Delivery Network (CDN) could soon stop working.

We contacted Cloudflare for comment but are still waiting for a response at the time of publication.

You might also like

• Russia to invest over half a billion US dollars in its battle against VPN usage
• Russia disconnects several regions to test its sovereign net
• Apple keeps removing VPN apps in Russia

• The FBI warns about web-based file conversion projects being malicious
• Some are dropping malware, others stealing sensitive data
• FBI urges victims to report the attacks

Free online file converters, joiners, and similar productivity tools are actually covers for data scraping and malware/ransomware distribution campaigns, the FBI is warning.

The Bureau's warning says its Denver Field Office agents are “increasingly seeing a scam involving free online document converter tools.”

In the warning, the law enforcement agency also urged victims to come forward and report instances of this scam.

Monitor your credit score with TransUnion starting at $29.95/month

TransUnion is a credit monitoring service that helps you stay on top of your financial health. With real-time alerts, credit score tracking, and identity theft protection, it ensures you never miss important changes. You'll benefit from a customizable online interface with clear insights into your credit profile. Businesses also benefit from TransUnion’s advanced risk assessment tools.

Preferred partner (What does this mean?)View Deal

Reporting the scam

“To conduct this scheme, cyber criminals across the globe are using any type of free document converter or downloader tool. This might be a website claiming to convert one type of file to another, such as a .doc file to a .pdf file. It might also claim to combine files, such as joining multiple .jpg files into one .pdf file. The suspect program might claim to be an MP3 or MP4 downloading tool.”

The worst part is that the tools work as intended, and the victims will still get the files they were looking for.

However, they will also get a piece of malware downloaded to their computer, or the files they uploaded for conversion/joining will be scrapped for information.

The FBI says the tools can scrape the submitted files for personal identifying information, such as social security numbers, dates of birth, and phone numbers, banking information, cryptocurrency information (seed phrases, wallet addresses, etc.), email addresses, and passwords.

“The best way to thwart these fraudsters is to educate people so they don’t fall victim to these fraudsters in the first place,” said FBI Denver Special Agent in Charge Mark Michalek.

“If you or someone you know has been affected by this scheme, we encourage you to make a report and take actions to protect your assets. Every day, we are working to hold these scammers accountable and provide victims with the resources they need.”

The FBI urges victims to report these scams to the FBI Internet Crime Complaint Center at www.ic3.gov.

You might also like

• This VPN is being abused to spread malware
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

Whether you're a shopper or a seller, a worker or a business owner, you likely have a lot on your mind when it comes to budgeting, planning trips or big purchases. And we want to hear all about it.

(Image credit: Spencer Platt)

• Microsoft is launching new AI agents for cybersecurity
• Thenew Copilot-powered agents will help automate the defense process
• Areas such as data security, identity management, and phishing could all get a boost

Microsoft is launching new Security Copilot Agents to help secure organizations with AI-first, end-to-end security platforms.

The company says its new agents are designed to “autonomously assist with critical areas” like data security, identity management, and phishing.

By working with some of the world’s top software companies, Microsoft hopes to deliver “game-changing” protections and help customers “scale, augment, and increase the effectiveness of their privacy operations” to help organizations navigate the increasingly complex threat landscape and regulatory requirements.

Core problems

Microsoft’s Global Head of Security, Vasu Jakkal, spoke to TechRadar Pro, to discuss the way that AI is changing the cybersecurity landscape, and how the new initiatives will help defenders use AI to their advantage.

Jakkal noted how AI is supercharging the volume of cyberattacks, and lowering the barriers for access to malicious campaigns, overwhelming security teams who often don’t have access to first-rate tools and rely on manual processes and ‘fragmented defenses’.

“So you look at these three core problems, threat landscape, operational complexity, and data security, there's no way humans can scale to keep up with these challenges. In fact, we don't have the human talent in security right now,” she warns.

To help security teams try and navigate this, Microsoft is unveiling 11 new Copilot agents.

Six of these agents will be available across the Microsoft end-to-end security platform, and are designed to assist with threat protection, data security, device management, identity and access, and threat intelligence.

The new launches come alongside Microsoft’s release of five new Agentic solutions to help bolster security teams worldwide.

These include a privacy breach response agent by OneTrust, a network supervisor agent by Aviatrix, a SecOps Tooling agent by BlueVoyant, an alert triage agent by Tanium, as well as a task optimizer agent by Fletch.

A helping hand

So that teams can keep up with the quickly evolving landscape, Security Copilot Agents will enable teams to handle high-volume security and IT tasks, and will work seamlessly alongside existing Microsoft security tools.

Microsoft Threat Intelligence now processes 84 trillion signals per day, revealing the exponential growth in cyber-attacks, including 7,000 password attacks per second.

Although you can’t ever eliminate the risk of human error entirely, these new tools will look to be a “another pair of eyes and pair of hands” to help double check things to reduce the risk factor, Jakkal explains.

“Last year, in one year, we saw 30 billion phishing emails. That's a lot. And this volume, you just can't keep up, humans can't triage these. And so the phishing agent now can triage these emails and alerts, and it can tell you, hey, this is a false alarm and this is a true alert, so it kind of reduces that volume”

The upper hand

Jakkal, like many others, describes cybersecurity as a cat and mouse game between cybercriminals and security teams.

Right now, AI is the attacker’s tool of choice and allows for a monumental number of intrusions, but the more attacks are leveraged, the more defenders can learn.

“Microsoft processes 84 trillion signals every single day. That signal intelligence, it's hard for humans to just work through that and scan through, but guess what tool works really great with data? AI.”

For security teams to gain the upper hand, defenders must embrace AI, Jakkal argues, as the talent gap and skills shortage is holding the industry back, and cybersecurity teams, “just don't have enough defenders in the world,” so must look to AI to keep up with demand.

The barrage of attacks isn’t likely to change anytime soon, either. Cyberattacks continue to be a profitable endeavour, and cybercrime is even helping fund rogue nations across the world, and with rising geopolitical tensions, cybersecurity teams must be more alert than ever before.

“Attacks are happening all around and because ransomware is a very lucrative industry and in fact global cybercrime costs us 9.2 trillion dollars, US dollars a year," Jakkal concludes.

"So as long as there's money to be made in it, we are going to see attacks and it can be even worse for a small and medium business because they don't have the staff to even tackle these problems.”

You might also like

• Take a look at our picks for the best AI tools around
• I've tested all the best AI agents including ChatGPT Deep Research and Gemini - these are the 5 top automated artificial intelligence tools you can try right now
• Microsoft - UK can help drive the global AI future, but only with the proper buy-in