When we look at the relationship between 73-year-old legendary football coach Bill Belichick and his girlfriend and business partner, 24-year-old Jordon Hudson, it’s hard to know exactly what we’re seeing. Two grown-ups in love forging a dynamic business partnership? Elder abuse, as some have wildly speculated? Or is it, as the vitriolic comments in Hudson’s social media posts would have it, good old-fashioned gold digging? Gold digging is a misogynistic and retro term, but this scandal is a retro one. It’s a bizarre, slightly off-putting mystery that fits remarkably well into our current age of newly regressive gender politics.Hudson and Belichick, the six-time Super Bowl-winning former NFL coach, met on a flight in 2021 and went public with their relationship last December. While their nearly 50-year age gap has raised eyebrows amid observers, the scandal has only grown as Hudson has taken on an increasingly central role in Belichick’s professional world. It’s a bizarre, slightly off-putting mystery that fits remarkably well into our current age of newly regressive gender politics.Belichick requested Hudson be cc’ed on all publicity and media emails about him at the University of North Carolina at Chapel Hill, where he is now a college football coach. In public records, she’s listed as the manager of several companies that appear to be tied to Belichick. Insider accounts say she essentially cast herself in a Dunkin Donuts commercial in which Belichick appeared, and that she blocked a docuseries about Belichick’s career. But speculation around the relationship reached a fever pitch when Hudson interrupted during Belichick’s interview on CBS Sunday Morning in April. “How did you guys meet?” host Tony Dokoupil asked Belichick, referring to Hudson. “We’re not talking about this,” Hudson cut in tersely from off-camera. “Jordon was a constant presence during our interview,” Dokoupil informs viewers in voiceover, in a moment that launched a thousand TikTok reaction videos. In the midst of this controversy, the couple has been manufacturing social media content about their relationship that isn’t alarming so much as it is lightly uncanny, especially given Belichick’s famously gruff public persona. On Hudson’s Instagram account, she has posted beachside pictures of herself balancing athletically on Belichick’s outstretched legs, and of Belichick, dressed as a fisherman, reeling in a mermaid-tailed Hudson from the surf. (“Ouucchhhh!!!” goes the caption on that one.)A few have defended the relationship between Belichick and Hudson. Sports media personality Colin Cowherd has said that Hudson’s choice to hop into Belichick’s interview was normal for PR directors, if “kind of a cringy thing.” Right-wing sports outlet The Outkick has rallied to Hudson’s side on the grounds that she triggers the libs, saying, “If Jordon wants to spend her weekends at Bill’s house on Nantucket, soaking up the sun and enjoying Cisco Brewery on Bill’s dime, I say GO FOR IT.” But the most common reaction to the spectacle of Hudson and Belichick’s relationship is the one outlined by sports media personality Katie Nolan on a podcast in February: This is weird and seems like it has sinister undertones.“We’re already going, ‘You’re how much younger than him?’ And then you show up in a commercial. And then you hear that she’s in charge in his career.” Nolan said. “It seems like you could be taking advantage of the guy. And he’s obviously taking advantage of the girl.” Notably, it doesn’t appear to be much of a head-scratcher why a powerful septuagenarian would choose to be with a recent college graduate and pageant contestant. But the nefarious reason implied about why Hudson wants to be with Belichick is the one people are misogynistically and freely throwing around: gold digger.The resurgence of the gold digger I don’t know anything about Hudson and Belichick’s private relationship, and most likely, neither do you. I don’t want to make any claims about who is using whom or what their private life is like. But as a feminist pop culture critic, I am interested to see the term “gold digger” swim back up from the collective unconsciousness again, ready and willing to go to work. There’s something almost old-fashioned about it. The last time “gold digger” was thrown around in pop culture so much was when teenager Courtney Stodden skyrocketed to fame in 2011 after they got married at age 16 to 51-year-old actor Doug Hutchison. Stodden would later describe their marriage as one characterized by grooming and sexual assault. Before that, the great pop culture gold digger was Anna Nicole Smith, the model turned paparazzi obsession who married oil tycoon J. Howard Marshall II in 1994. Since Smith’s untimely death in 2007, she’s found a place in the pantheon of wronged women of the 1990s, someone we came to believe over the last decade was publicly mistreated — in part by her tarring as a gold digger. “It’s a bit provincial to look down one’s nose at a woman wanting something from a man and give a pass to a man who arguably would have never given her a second glance had she not been a beautiful, blonde Playboy Playmate,” essayist and sex work activist Laura LeMoon argued in 2023, writing about Smith in Salon. “Rather than singling out and condemning Anna Nicole Smith, we should be pointing our fingers at the inequities and systemic failures that put people like me and Smith in positions where obtaining money and resources from men, directly or indirectly, is our best option for survival.”“Gold digger” is a euphemism that allows us as a culture to talk around those systemic failures. It’s a way to deride women when they take seriously the idea that their financial well-being should depend on their relationships with men. So it’s odd to see the term becoming so popular during a moment when popular culture has become rather infatuated with the idea that life is most pleasant, simple, and straightforward when women’s finances do depend on romantic relationships with men.TikTok is full of tradwives explaining how their lives became better once they got out of the 9-to-5 grind to make cereal from scratch for their children and let their husbands do the breadwinning. (That the most successful of the tradwife influencers are actually the primary breadwinners in their homes is an irony the influencers themselves seldom discuss.) Or there are the stay-at-home girlfriends cooing over how taking care of their boyfriends without even the financial safety net of a marriage contract has been their ticket to the soft life. The gold digger is the tradwife as seen through a funhouse mirror: a woman living off her ability to attract a man, only here done crassly.Meanwhile, some of President Donald Trump’s most vocal supporters are arguing that his tariffs will force women out of the workforce and make them once again financially dependent on men, either through marriage or through sex work, life paths these men treat as equivalent. Trump’s appeal to the dream of a lost American manufacturing economy is a nostalgic one. It’s an appealing fantasy to some of his more incel-adjacent fans: that under this economy, the sexes will revert to an older, allegedly more natural economic relationship, one in which women trade their sexuality and childrearing capabilities to men in exchange for financial security. The gold digger is the tradwife as seen through a funhouse mirror: a woman living off her ability to attract a man, only here done crassly, without the show of love to soften the crude edges of the transaction. The gold digger is simply a figure we can blame for how uncomfortable this dynamic makes us feel, without having to think through just what is so uncomfortable about it. So if we’re angry at Jordon Hudson, it’s worth asking the question: Are we angry with her, or with the fact that powerful people want to make gold digging one of a woman’s most viable career paths again?See More:

The cobbled streets and centuries-old university halls of Leuven recently served as a picturesque backdrop for the Materialise 3D Printing in Hospitals Forum 2025. Belgium’s Flemish Brabant capital hosted the annual meeting, which has become a key gathering for the medical 3D printing community since its launch in 2017. This year, 140 international healthcare professionals convened for two days of talks, workshops, and lively discussion on how Materialise’s software enhances patient care. The Forum’s opening day, hosted at Leuven’s historic Irish College, featured 16 presentations by 18 healthcare clinicians and medical 3D printing experts.  While often described as the future of medicine, personalized healthcare has already become routine in many clinical settings. Speakers emphasized that 3D printing is no longer merely a “cool” innovation, but an essential tool that improves patient outcomes. “Personalized treatment is not just a vision for the future,” said Koen Peters, Executive Vice President Medical at Materialise. “It’s a reality we’re building together every day.” During the forum, practitioners and clinical engineers demonstrated the critical role of Materialise’s software in medical workflows. Presentations highlighted value across a wide range of procedures, from brain tumour removal and organ transplantation to the separation of conjoined twins and maxillofacial implant surgeries. Several use cases demonstrated how 3D technology can reduce surgery times by up to four times, enhance patient recovery, and cut hospital costs by almost £6,000 per case.      140 visitors attended the Materialise 3D Printing in Hospitals Forum 2025. Photo via Materialise. Digital simulation and 3D printing slash operating times  Headquartered a few miles outside Leuven’s medieval center, Materialise is a global leader in medical 3D printing and digital planning. Its Mimics software suite automatically converts CT and MRI scans into detailed 3D models. Clinicians use these tools to prepare for procedures, analyse anatomy, and create patient-specific models that enhance surgical planning. So far, Materialise software has supported more than 500,000 patients and analysed over 6 million medical scans. One case that generated notable interest among the Forum’s attendees was that of Lisa Ferrie and Jiten Parmar from Leeds General Infirmary. The pair worked alongside Asim Sheikh, a Consultant Skullbase and Neurovascular Neurosurgeon, to conduct the UK’s first “coach door osteotomy” on Ruvimbo Kaviya, a 40-year-old nurse from Leeds.  This novel keyhole surgery successfully removed a brain tumor from Kaviya’s cavernous sinus, a hard-to-reach area behind the eyes. Most surgeries of this kind require large incisions and the removal of substantial skull sections, resulting in extended recovery time and the risk of postoperative complications. Such an approach would have presented serious risks for removing Kaviya’s tumor, which “was in a complex area surrounded by a lot of nerves,” explained Parmar, a Consultant in Maxillofacial Surgery.    Instead, the Leeds-based team uses a minimally invasive technique that requires only a 1.5 cm incision near the side of Ravimbo’s eyelid. A small section of skull bone was then shifted sideways and backward, much like a coach door sliding open, to create an access point for tumor removal. Following the procedure, Ravimbo recovered in a matter of days and was left with only a 6 mm scar at the incision point.  Materialise software played a vital role in facilitating this novel procedure. Ferrie is a Biomedical Engineer and 3D Planning Service Lead at Leeds Teaching Hospitals NHS Trust. She used mimics to convert medical scans into digital 3D models of Ravimbo’s skull. This allowed her team to conduct “virtual surgical planning” and practice the procedure in three dimensions, “to see if it’s going to work as we expect.”  Ferrie also fabricated life-sized, polyjet 3D printed anatomical models of Ravimbo’s skull for more hands-on surgical preparation. Sheikh and Parmar used these models in the hospital’s cadaver lab to rehearse the procedure until they were confident of a successful outcome. This 3D printing-enabled approach has since been repeated for additional cases, unlocking a new standard of care for patients with previously inoperable brain tumors.  The impact of 3D planning is striking. Average operating times fell from 8-12 hours to just 2-3 hours, and average patient discharge times dropped from 7-10 days to 2-3 days. These efficiencies translated into cost savings of £1,780 to £5,758 per case, while additional surgical capacity generated an average of £11,226 in income per operating list. Jiten Parmar (right) and Lisa Ferrie (left) presenting at the Materialise 3D Printing in Hospitals Forum 2025. Photo via Materialise. Dr. Davide Curione also discussed the value of virtual planning and 3D printing for surgical procedures. Based at Bambino Gesù Pediatric Hospital in Rome, the radiologist’s team conducts 3D modeling, visualization, simulation, and 3D printing.  One case involved thoraco-omphalopagus twins joined at the chest and abdomen. Curione’s team 3D printed a multi-color anatomical model of the twins’ anatomy, which he called “the first of its kind for complexity in Italy.” Fabricated in transparent resin, the model offered a detailed view of the twins’ internal anatomy, including the rib cage, lungs, and cardiovascular system. Attention then turned to the liver. The team built a digital reconstruction to simulate the optimal resection planes for the general separation and the hepatic splitting procedure. This was followed by a second multi-colour 3D printed model highlighting the organ’s vascularisation. These resources improved surgical planning, cutting operating time by 30%, and enabled a successful separation, with no major complications reported two years post-operation. Dr. Davide Curione’s workflow for creating a 3D printed model of thoraco-omphalopagus twins using Mimics. Image via Frontiers in Physiology. VR-enabled surgery enhances organ transplants   Materialise’s Mimics software can also be used in extended reality (XR), allowing clinicians to interact more intuitively with 3D anatomical models and medical images. By using off-the-shelf virtual reality (VR) and augmented reality (AR) headsets, healthcare professionals can more closely examine complex structures in an immersive environment. Dr. David Sibřina is a Principal Researcher and Developer for the VRLab team at Prague’s Institute for Clinical and Experimental Medicine (IKEM). He leads efforts to accelerate the clinical adoption of VR and AR in organ transplantation, surgical planning, and surgical guidance.  The former Forbes 30 Under 30 honouree explained that since 2016, IKEM’s 3D printing lab has focused on producing anatomical models to support liver and kidney donor programmes. His lab also fabricates 3D printed anatomical models of ventricles and aneurysms for clinical use.  However, Sibřina’s team recently became overwhelmed by high demand for physical models, with surgeons requesting additional 3D model processing options. This led Sibřina to create the IKEM VRLab, offering XR capabilities to help surgeons plan and conduct complex transplantation surgeries and resection procedures.      When turning to XR, Sibřina’s lab opted against adopting a ready-made software solution, instead developing its own from scratch. “The problem with some of the commercial solutions is capability and integration,” he explained. “The devices are incredibly difficult and expensive to integrate within medical systems, particularly in public hospitals.” He also pointed to user interface shortcomings and the lack of alignment with established medical protocols.  According to Sibřina, IKEM VRLab’s offering is a versatile and scalable VR system that is simple to use and customizable to different surgical disciplines. He described it as “Zoom for 3D planning,” enabling live virtual collaboration between medical professionals. It leverages joint CT and MRI acquisition models, developed with IKEM’s medical physicists and radiologists. Data from patient scans is converted into interactive digital reconstructions that can be leveraged for analysis and surgical planning.  IKEM VRLab also offers a virtual “Fitting Room,” which allows surgeons to assess whether a donor’s organ size matches the recipient’s body. A digital model is created for every deceased donor and live recipient’s body, enabling surgeons to perform the size allocation assessments.  Sibřina explained that this capability significantly reduces the number of recipients who would otherwise fail to be matched with a suitable donor. For example, 262 deceased liver donors have been processed for Fitting Room size allocations by IKEM VRLab. In 27 instances, the VR Fitting Room prevented potential recipients from being skipped in the waiting list based on standard biometrics, CT axis measurements, and BMI ratios.                          Overall, 941 patient-specific visualizations have been performed using Sibřina’s technology. 285 (28%) were for liver recipients, 311 (31%) for liver donors, and 299 (23%) for liver resection. Living liver donors account for 59 (6%) cases, and split/reduced donors for 21 (2%).           A forum attendee using Materialise’s Mimics software in augmented reality (AR). Photo via Materialise. Personalized healthcare: 3D printing implants and surgical guides  Beyond surgical planning and 3D visualisation, Materialise Mimics software supports the design and production of patient-specific implants and surgical guides. The company conducts healthcare contract manufacturing at its Leuven HQ and medical 3D printing facility in Plymouth, Michigan.  Hospitals can design patient-specific medical devices in-house or collaborate with Materialise’s clinical engineers to develop custom components. Materialise then 3D prints these devices and ships them for clinical use. The Belgian company, headed by CEO Brigitte de Vet-Veithen, produces around 280,000 custom medical instruments each year, with 160,000 destined for the US market. These include personalised titanium cranio-maxillofacial (CMF) implants for facial reconstruction and colour-coded surgical guides. Poole Hospital’s 3D specialists, Sian Campbell and Poppy Taylor-Crawford, shared how their team has adopted Materialise software to support complex CMF surgeries. Since acquiring the platform in 2022, they have developed digital workflows for planning and 3D printing patient-specific implants and surgical guides in 14 cases, particularly for facial reconstruction.  Campbell and Taylor-Crawford begin their workflow by importing patient CT and MRI data into Materialise’s Mimics Enlight CMF software. Automated tools handle initial segmentation, tumour resection planning, and the creation of cutting planes. For more complex cases involving fibula or scapula grafts, the team adapts these workflows to ensure precise alignment and fit of the bone graft within the defect. Next, the surgical plan and anatomical data are transferred to Materialise 3-matic, where the team designs patient-specific resection guides, reconstruction plates, and implants. These designs are refined through close collaboration with surgeons, incorporating feedback to optimise geometry and fit. Virtual fit checks verify guide accuracy, while further analysis ensures compatibility with surgical instruments and operating constraints. Once validated, the guides and implants are 3D printed for surgery. According to Campbell and Taylor-Crawford, these custom devices enable more accurate resections and implant placements. This improves surgical alignment and reduces theatre time by minimising intraoperative adjustments. An example of the cranio-maxillofacial implants and surgical guides 3D printed by Materialise. Photo by 3D Printing Industry Custom 3D printed implants are also fabricated at the Rizzoli Orthopaedic Institute in Bologna, Italy. Originally established as a motion analysis lab, the institute has expanded its expertise into surgical planning, biomechanical analysis, and now, personalized 3D printed implant design. Dr. Alberto Leardini, Director of the Movement Analysis Laboratory, described his team’s patient-specific implant workflow. They combine CT and MRI scans to identify bone defects and tumour locations. Clinical engineers then use this data to build digital models and plan resections. They also design cutting guides and custom implants tailored to each patient’s anatomy. These designs are refined in collaboration with surgeons before being outsourced to manufacturing partners for production. Importantly, this workflow internalizes design and planning phases. By hosting engineering and clinical teams together on-site, they aim to streamline decision-making and reduce lead times. Once the digital design is finalised, only the additive manufacturing step is outsourced, ensuring “zero distance” collaboration between teams.  Dr. Leardini emphasised that this approach improves clinical outcomes and promises economic benefits. While custom implants require more imaging and upfront planning, they reduce time in the operating theatre, shorten hospital stays, and minimise patient transfers.  After a full day of presentations inside the Irish College’s eighteenth-century chapel, the consensus was clear. 3D technology is not a niche capability reserved for high-end procedures, but a valuable tool enhancing everyday care for thousands of patients globally. From faster surgeries to cost savings and personalized treatments, hospitals are increasingly embedding 3D technology into routine care. Materialise’s software sits at the heart of this shift, enabling clinicians to deliver safer, smarter, and more efficient healthcare.  Take the 3DPI Reader Survey – shape the future of AM reporting in under 5 minutes. Read all the 3D printing news from RAPID + TCT 2025 Subscribe to the 3D Printing Industry newsletter to keep up with the latest 3D printing news.You can also follow us on LinkedIn, and subscribe to the 3D Printing Industry Youtube channel to access more exclusive content.Featured image shows 3D printed anatomical models at Materialise HQ in Leuven. Photo by 3D Printing Industry.

Captured on Nintendo Switch (Handheld/Undocked)It's possibly becoming a bit of a hard sell these days, the old roguelike dungeon-crawling adventure, replete with retro/pixel visuals. We know it. However, as busy and as overstuffed with pale imitators to the greats as this genre is, it's still well worth keeping an eye out for the odd gem. And here we have an example of said gem. Into the Restless Ruins is a fantastic thing - put simply, a game that takes a whole bunch of mechanics you might think you're tired of, and weaves them together into something entirely fresh and unique. Edinburgh-based Ant Workshop Studios has melded deckbuilding, dungeon-crawling, dungeon-building, a fiendishly moreish memory game, and delightful, Vampire Survivors-esque, auto-combat to beguiling effect in its Switch debut. The delicately judged balance here, in how each of these mechanics is given an equal share of your time, ensures that, from the first dungeon drop to the very last battle of the campaign, you may find yourself immobilised, such is the addictiveness of the core gameplay loops running in tandem. Captured on Nintendo Switch (Handheld/Undocked) Kicking off in a dank catacomb, in a region known as Eorisdale, Into the Restless Ruins sees you take charge of a cloaked protagonist, a Celtic champion, who's on the hunt for the Harvest Maiden. This figure from Scottish folklore, according to the game, can satisfy any heart's greatest desires - in return for a little toil. And so off we pop down some dungeons as we've got a hot tip-off that's where she goes to chill out. This first dungeon gently introduces the flow of the gameplay. Starting a run, you're procedurally-generated a single piece of dungeon, perhaps a crossroads, with a spawn point in the centre. You also have a deck with five cards. Each card has a little part of a dungeon displayed on it, alongside the number of BP (Build Points) you'll need to play it, and a note of any other effects the card may have. An armoury will raise your attack stat, a camp lets you regenerate a little of your torch, whilst faded groves give you a little health regen, and so on through more complex fare. Captured on Nintendo Switch (Docked) You start off from your spawn point and use your cards, choosing a piece of dungeon to fit — in a lovely clicky way — to the piece you're currently on. The aim now is to build and expand, using your given turns and BP, to reach several clouded areas spread around you, some of which contain seals. These seals must be found and broken to advance further to a final boss confrontation. All good. Now, once you've done your planning and played your cards, you hold in 'Y' to drop right down into the dungeon (and it's nice how the game just seamlessly zooms right in for this bit) where the auto-combat battling begins. The combat here will come as no surprise to anyone who's survived vampires: it's all about positioning and timing, avoiding explodey rats and mice, weaving through wizard's spells, and knowing when to press or pull back. Once this part starts, you'll likely quickly realise that you've made some mistakes in the planning (especially if you are us). You see, as your dungeons expand, as you reach a seal and break it and are then faced down by a horde of angry beasties, the importance of dungeon design is fully revealed. Accessibility Features GameplayAssistsControlsVisualAudioFull Accessibility Report Data by Family Gaming Database Campsites, first up, need to be placed along routes with some thought to ensure you can keep your torch lit, because they've only gone and put your light supply on a tightly-timed gauge. Tension has entered the chat! You also have a bar dictating your health, which doesn't take much to empty either, so the positioning of those regenerating faded groves is now of critical import. Once you complete a run you'll be taken back to your deck and another gauge is deployed. This purple addition to the fun shows you how cursed you are, you see. It gets added to with each complete run, and so doing well, not dying, and playing clever will help you keep this bar filling up as slowly as possible. Dying and making a mess, on the other hand, will add some hefty chunks, putting you at risk of being overwhelmed before you manage to break all the seals and beat the boss. Phew! The first dungeon's leisurely pace and straightforward layout gives you a chance to get used to spinning and placing parts of your dungeon onto the map, as well as getting to grips with how you'll need to time everything just so to push on to victory. The combat keeps itself varied by providing new weapons and collectibles, too, as well as giving you new options through a vast selection of upgradeable and changeable cards (116 base varieties, we believe). Captured on Nintendo Switch (Handheld/Undocked) There are further layers provided through build synergies — certain placement patterns provide surprise boons — and there's also a whole bunch of Cantrips to unlock and use, allowing you to add both positive and negative effects to raise or lower the amount of XP you gain in a run, or "harvest" to use the correct lingo. So, lots of moving parts to consider. You've also got a few characters to meet out in these ruins that you've built around yourself, strange NPCs who'll upgrade your cards, duplicate particular faves, and more besides, whilst imparting some folklore upon you. Upgrading here, it has to be said, is also just really cool and satisfying because, besides the usual numbers-going-up, you get new entrances onto your card pieces. So, for example, a long corridor with three doors that you couldn't use because it wouldn't fit, now gets an extra door for you to work with. It's addictive, puzzley-time stuff that just gets better as it gets harder. Captured on Nintendo Switch (Docked) Of course, when a clouded area within any of your dungeons doesn't contain a seal, you've wasted a lot of resources making your way towards it, and this aspect also brings enough tension to your planning time that the building is every bit as enjoyable as the combat. Wandering through dungeons brings the added danger of getting lost, too, as there's no map, so the memory game starts to rear its head as you move onto the later catacombs, which mix things up by spreading the clouds you need to uncover in various directions. This means way more pre-planning is required to avoid coming a cropper to the tight time and health constraints placed on you. Beyond this, you've also got cursed cards that'll be played into your deck once you've been cursed, these add some rough negatives to proceedings if you let them sit in your hand, and so you need to consider when to play them (they cost a BP) in order to do the least amount of damage to your chances and get them out of your deck. These are offset by favour cards you'll get from breaking seals and levelling up, so you could, as an example, play a cursed card that means you can't heal up at groves, but then play a card that gives you +500% damage, so it nullifies the negative somewhat and removes it from your deck. Captured on Nintendo Switch (Docked) There are so many wrinkles here that it could have become very tedious and/or messy, so the best thing is how it's all so effortless and easy to learn, even if you don't really do cards as a regular thing. Every aspect is explained well (there's a tutorial in the menus when you need it), and you can take your time in the planning phase. As the catacombs get harder, things progress along perfectly from relaxing and addictive, to tension-filled and even more bloody addictive. The lovely pixel art style, quick shots of Scottish folklore, and solid soundtrack all work to give you exactly the kind of ambience you need to settle in for the long haul. In terms of negatives, well, the story aspects are light, really, so this one isn't gonna be challenging any of the more narratively bombastic examples of the genre to a straight-up auto-fight in the school playground. We guess, too, that the graphical style is one that's been done quite a bit, but honestly, we're picking for the sake of it now. This is some premium-grade roguelike, in the end - a uniquely fresh take that borrows inspirations from all over its chosen genre, and cobbles them together into a wonderful indie game that's very much its own thing. Just be warned that the 'one more run' factor here is high. Conclusion Into The Restless Ruins has taken us totally by surprise. This is a belting addition to the Switch's lineup of roguelikes that manages to make itself fresh and unique in a category full of bland copycats. The melding of deckbuilding, combat, memory challenges, and construction of dungeons makes for a game that draws you right in every time, robbing you of hours as you work your way through its delightful catacombs. A little more narrative oomph may have taken it even further, but you're still looking at a bit of a must-play here. A wonderfully clever melding of mechanicsCombat is fun with good variety in enemies and weaponsPacing is spot on, with dungeons getting more complex as you goEasy to get to grips with but deep as the darkest dungeon, too Story is very lightPixel art style has been done quite a bit before Excellent 9/10

May 16, 2025Ravie LakshmananMalware / Cyber Attack Cybersecurity researchers have shed light on a new malware campaign that makes use of a PowerShell-based shellcode loader to deploy a remote access trojan called Remcos RAT. "Threat actors delivered malicious LNK files embedded within ZIP archives, often disguised as Office documents," Qualys security researcher Akshay Thorve said in a technical report. "The attack chain leverages mshta.exe for proxy execution during the initial stage." The latest wave of attacks, as detailed by Qualys, employs tax-related lures to entice users into opening a malicious ZIP archive containing a Windows shortcut (LNK) file, which, in turn, makes use of mshta.exe, a legitimate Microsoft tool used to run HTML Applications (HTA). The binary is used to execute an obfuscated HTA file named "xlab22.hta" hosted on a remote server, which incorporates Visual Basic Script code to download a PowerShell script, a decoy PDF, and another HTA file similar to xlab22.hta called "311.hta." The HTA file is also configured to make Windows Registry modifications to ensure that "311.hta" is automatically launched upon system startup. Once the PowerShell script is executed, it decodes and reconstructs a shellcode loader that ultimately proceeds to launch the Remcos RAT payload entirely in memory. Remcos RAT is a well-known malware that offers threat actors full control over compromised systems, making it an ideal tool for cyber espionage and data theft. A 32-bit binary compiled using Visual Studio C++ 8, it features a modular structure and can gather system metadata, log keystrokes, capture screenshots, monitor clipboard data, and retrieve a list of all installed programs and running processes. In addition, it establishes a TLS connection to a command-and-control (C2) server at "readysteaurants[.]com," maintaining a persistent channel for data exfiltration and control. This is not the first time fileless versions of Remcos RAT have been spotted in the wild. In November 2024, Fortinet FortiGuard Labs detailed a phishing campaign that filelessly deployed the malware by making use of order-themed lures. What makes the attack method attractive to threat actors is that it allows them to operate undetected by many traditional security solutions as the malicious code runs directly in the computer's memory, leaving very few traces on the disk. "The rise of PowerShell-based attacks like the new Remcos RAT variant demonstrates how threat actors are evolving to evade traditional security measures," J Stephen Kowski, Field CTO at SlashNext, said. "This fileless malware operates directly in memory, using LNK files and MSHTA.exe to execute obfuscated PowerShell scripts that can bypass conventional defenses. Advanced email security that can detect and block malicious LNK attachments before they reach users is crucial, as is real-time scanning of PowerShell commands for suspicious behaviors." The disclosure comes as Palo Alto Networks Unit 42 and Threatray detailed a new .NET loader that's used to detonate a wide range of commodity information stealers and RATS like Agent Tesla, NovaStealer, Remcos RAT, VIPKeylogger, XLoader, and XWorm. The loader features three stages that work in tandem to deploy the final-stage payload: A .NET executable that embeds the second and third stages in encrypted form, a .NET DLL that decrypts and loads the next stage, and a .NET DLL that manages the deployment of the main malware. "While earlier versions embedded the second stage as a hardcoded string, more recent versions use a bitmap resource," Threatray said. "The first stage extracts and decrypts this data, then executes it in memory to launch the second stage." Unit 42 described the use of bitmap resources to conceal malicious payloads a a steganography technique that can bypass traditional security mechanisms and evade detection. The findings also coincide with the emergence of several phishing and social engineering campaigns that are engineered for credential theft and malware delivery - Use of trojanized versions of the KeePass password management software – codenamed KeeLoader – to drop a Cobalt Strike beacon and steal sensitive KeePass database data, including administrative credentials. The malicious installers are hosted on KeePass typosquat domains that are served via Bing ads. Use of ClickFix lures and URLs embedded within PDF documents and a series of intermediary dropper URLs to deploy Lumma Stealer. Use of booby-trapped Microsoft Office documents that are used to deploy the Formbook information stealer protected using a malware distribution service referred to as Horus Protector. Use of blob URIs to locally loads a credential phishing page via phishing emails, with the blob URIs served using allow-listed pages (e.g., onedrive.live[.]com) that are abused to redirect victims to a malicious site that contains a link to a threat actor-controlled HTML page. Use of RAR archives masquerading as setup files to distribute NetSupport RAT in attacks targeting Ukraine and Poland. Use of phishing emails to distribute HTML attachments that contain malicious code to capture victims' Outlook, Hotmail, and Gmail credentials and exfiltrate them to a Telegram bot named "Blessed logs" that has been active since February 2025 The developments have also been complemented by the rise in artificial intelligence (AI)-powered campaigns that leverage polymorphic tricks that mutate in real-time to sidestep detection efforts. These include modifying email subject lines, sender names, and body content to slip past signature-based detection. "AI gave threat actors the power to automate malware development, scale attacks across industries, and personalize phishing messages with surgical precision," Cofense said. "These evolving threats are increasingly able to bypass traditional email filters, highlighting the failure of perimeter-only defenses and the need for post-delivery detection. It also enabled them to outmaneuver traditional defenses through polymorphic phishing campaigns that shift content on the fly. The result: deceptive messages that are increasingly difficult to detect and even harder to stop." Found this article interesting? Follow us on Twitter  and LinkedIn to read more exclusive content we post. SHARE    

This Eye-Catching Rattlesnake Found in Arizona Has Unusual ‘Leopard Spots,’ Likely From a Genetic Mutation Snake wranglers safely relocated the healthy, female western diamondback from a backyard in Scottsdale, but they say the find is a first in their experience The healthy adult female was covered in curvy circles, similar to leopard spots. Rattlesnake Solutions / Facebook Arizonans are accustomed to seeing western diamondbacks slithering around or sunning themselves on rocks. But recently, rattlesnake wranglers came across a “weird” and “funky-looking” individual with highly unusual patterning. Experts with Rattlesnake Solutions, a humane snake relocation and prevention company, were called in to remove a western diamondback (Crotalus atrox) from a backyard in Scottsdale on May 9. When they arrived, they found a 30-inch-long, healthy, adult female with a “big pattern mutation,” according to a post the company shared on Facebook. “In the many thousands of diamondbacks we’ve seen over the years, this is a first,” the company adds. Instead of the typical, diamond-shaped pattern these snakes are named for, the female was covered in round, curvy spots. The snake’s coloring was also uncommon. “It’s hard to tell exactly what is going on here, but it looks like the normal diamondback pattern is disrupted,” Bryan Hughes, the owner of Rattlesnake Solutions, tells McClatchy News’ Mark Price in an email. “The resulting pattern looks like circles, or ‘leopard spots’ as some have noted. It also lacks the characteristic facial stripes, has a unique pattern on the head and has peachy orange tones throughout that are atypical.” Near the snake’s rattle at the end of its tail is a section of black and white patterning. “That’s the only normal thing about it,” Rattlesnake Solutions writes in a comment on Facebook. Western diamondbacks typically have evenly sized bands of alternating white and black close to the rattle.Based on the location, Hughes does not believe the snake is a hybrid, or the offspring of two different parent species. Instead, he suspects the odd pattern is the result of “a unique mutation in the wild,” he tells McClatchy News. If that’s the case, Hughes worries the mutation could be a symptom of “environmental stress.” “This is a very pretty snake, but it may also be a warning,” he tells McClatchy News. “The area where this snake was found is in a zone of heavy urban development and increasing yearly temperatures. … If mutations like this are an indicator of genetic mutation in response to a warming climate, it may be something we see more of in urban areas.” Speaking with People magazine’s Madison E. Goldberg, Hughes says he wonders whether the pattern is hereditary or if the state’s recent dry spell could have affected the snake. Arizona has been experiencing a “severe and sustained” drought since the mid- to late 1990s, according to the Arizona Department of Water Resources. More recently, the period between April 2020 and March 2025 was the third warmest and fourth driest on record. In 2024, the Phoenix metropolitan area—which includes Scottsdale—had its hottest year ever, breaking the previous record by 1.3 degrees. The area saw 70 days with high temperatures at or above 110 degrees, compared to the typical 21 days. Though the leopard-like pattern is extremely rare, snake aficionados do occasionally find wild serpents with mutations. Typically, however, the mutations show up as stripes or missing patterns. “In the past 25 years, only three snakes that have a similar mutation have been seen: this one, another in the same general area in 2016 and the first found around 20 years ago,” Hughes tells People magazine. “It’s the first I’ve ever seen, and likely the last.” After the Rattlesnake Solutions team captured the snake in the backyard in Scottsdale, they moved it to another location and released it back into the wild. Rattlesnake season is well underway in Arizona, which is home to around 13 species—including the Arizona ridge-nosed rattlesnake (Crotalus willardi), the official state reptile. Rattlesnakes tend to be most active from March through October. Get the latest stories in your inbox every weekday.

I’ve been working as a data science consultant for the past three years, and I’ve had the opportunity to work on multiple projects across various industries. Yet, I noticed one common denominator among most of the clients I worked with: They rarely have a clear idea of the project objective. This is one of the main obstacles data scientists face, especially now that Gen AI is taking over every domain. But let’s suppose that after some back and forth, the objective becomes clear. We managed to pin down a specific question to answer. For example: I want to classify my customers into two groups according to their probability to churn: “high likelihood to churn” and “low likelihood to churn” Well, now what? Easy, let’s start building some models! Wrong! If having a clear objective is rare, having a reliable benchmark is even rarer. In my opinion, one of the most important steps in delivering a data science project is defining and agreeing on a set of benchmarks with the client. In this blog post, I’ll explain: What a benchmark is, Why it is important to have a benchmark, How I would build one using an example scenario and Some potential drawbacks to keep in mind What is a benchmark? A benchmark is a standardized way to evaluate the performance of a model. It provides a reference point against which new models can be compared. A benchmark needs two key components to be considered complete: A set of metrics to evaluate the performance A set of simple models to use as baselines The concept at its core is simple: every time I develop a new model I compare it against both previous versions and the baseline models. This ensures improvements are real and tracked. It is essential to understand that this baseline shouldn’t be model or dataset-specific, but rather business-case-specific. It should be a general benchmark for a given business case. If I encounter a new dataset, with the same business objective, this benchmark should be a reliable reference point. Why building a benchmark is important Now that we’ve defined what a benchmark is, let’s dive into why I believe it’s worth spending an extra project week on the development of a strong benchmark. Without a Benchmark you’re aiming for perfection — If you are working without a clear reference point any result will lose meaning. “My model has a MAE of 30.000” Is that good? IDK! Maybe with a simple mean you would get a MAE of 25.000. By comparing your model to a baseline, you can measure both performance and improvement. Improves Communicating with Clients — Clients and business teams might not immediately understand the standard output of a model. However, by engaging them with simple baselines from the start, it becomes easier to demonstrate improvements later. In many cases benchmarks could come directly from the business in different shapes or forms. Helps in Model Selection — A benchmark gives a starting point to compare multiple models fairly. Without it, you might waste time testing models that aren’t worth considering. Model Drift Detection and Monitoring — Models can degrade over time. By having a benchmark you might be able to intercept drifts early by comparing new model outputs against past benchmarks and baselines. Consistency Between Different Datasets — Datasets evolve. By having a fixed set of metrics and models you ensure that performance comparisons remain valid over time. With a clear benchmark, every step in the model development will provide immediate feedback, making the whole process more intentional and data-driven. How I would build a benchmark I hope I’ve convinced you of the importance of having a benchmark. Now, let’s actually build one. Let’s start from the business question we presented at the very beginning of this blog post: I want to classify my customers into two groups according to their probability to churn: “high likelihood to churn” and “low likelihood to churn” For simplicity, I’ll assume no additional business constraints, but in real-world scenarios, constraints often exist. For this example, I am using this dataset (CC0: Public Domain). The data contains some attributes from a company’s customer base (e.g., age, sex, number of products, …) along with their churn status. Now that we have something to work on let’s build the benchmark: 1. Defining the metrics We are dealing with a churn use case, in particular, this is a binary classification problem. Thus the main metrics that we could use are: Precision — Percentage of correctly predicted churners among all predicted churners Recall — Percentage of actual churners correctly identified F1 score — Balances precision and recall True Positives, False Positives, True Negative and False Negatives These are some of the “simple” metrics that could be used to evaluate the output of a model. However, it is not an exhaustive list, standard metrics aren’t always enough. In many use cases, it might be useful to build custom metrics. Let’s assume that in our business case the customers labeled as “high likelihood to churn” are offered a discount. This creates: A cost ($250) when offering the discount to a non-churning customer A profit ($1000) when retaining a churning customer Following on this definition we can build a custom metric that will be crucial in our scenario: # Defining the business case-specific reference metric def financial_gain(y_true, y_pred): loss_from_fp = np.sum(np.logical_and(y_pred == 1, y_true == 0)) * 250 gain_from_tp = np.sum(np.logical_and(y_pred == 1, y_true == 1)) * 1000 return gain_from_tp - loss_from_fp When you are building business-driven metrics these are usually the most relevant. Such metrics could take any shape or form: Financial goals, minimum requirements, percentage of coverage and more. 2. Defining the benchmarks Now that we’ve defined our metrics, we can define a set of baseline models to be used as a reference. In this phase, you should define a list of simple-to-implement model in their simplest possible setup. There is no reason at this state to spend time and resources on the optimization of these models, my mindset is: If I had 15 minutes, how would I implement this model? In later phases of the model, you can add mode baseline models as the project proceeds. In this case, I will use the following models: Random Model — Assigns labels randomly Majority Model — Always predicts the most frequent class Simple XGB Simple KNN import numpy as np import xgboost as xgb from sklearn.neighbors import KNeighborsClassifier class BinaryMean(): @staticmethod def run_benchmark(df_train, df_test): np.random.seed(21) return np.random.choice(a=[1, 0], size=len(df_test), p=[df_train['y'].mean(), 1 - df_train['y'].mean()]) class SimpleXbg(): @staticmethod def run_benchmark(df_train, df_test): model = xgb.XGBClassifier() model.fit(df_train.select_dtypes(include=np.number).drop(columns='y'), df_train['y']) return model.predict(df_test.select_dtypes(include=np.number).drop(columns='y')) class MajorityClass(): @staticmethod def run_benchmark(df_train, df_test): majority_class = df_train['y'].mode()[0] return np.full(len(df_test), majority_class) class SimpleKNN(): @staticmethod def run_benchmark(df_train, df_test): model = KNeighborsClassifier() model.fit(df_train.select_dtypes(include=np.number).drop(columns='y'), df_train['y']) return model.predict(df_test.select_dtypes(include=np.number).drop(columns='y')) Again, as in the case of the metrics, we can build custom benchmarks. Let’s assume that in our business case the the marketing team contacts every client who’s: Over 50 y/o and That is not active anymore Following this rule we can build this model: # Defining the business case-specific benchmark class BusinessBenchmark(): @staticmethod def run_benchmark(df_train, df_test): df = df_test.copy() df.loc[:,'y_hat'] = 0 df.loc[(df['IsActiveMember'] == 0) & (df['Age'] >= 50), 'y_hat'] = 1 return df['y_hat'] Running the benchmark To run the benchmark I will use the following class. The entry point is the method compare_with_benchmark() that, given a prediction, runs all the models and calculates all the metrics. import numpy as np class ChurnBinaryBenchmark(): def __init__( self, metrics = [], benchmark_models = [], ): self.metrics = metrics self.benchmark_models = benchmark_models def compare_pred_with_benchmark( self, df_train, df_test, my_predictions, ): output_metrics = { 'Prediction': self._calculate_metrics(df_test['y'], my_predictions) } dct_benchmarks = {} for model in self.benchmark_models: dct_benchmarks[model.__name__] = model.run_benchmark(df_train = df_train, df_test = df_test) output_metrics[f'Benchmark - {model.__name__}'] = self._calculate_metrics(df_test['y'], dct_benchmarks[model.__name__]) return output_metrics def _calculate_metrics(self, y_true, y_pred): return {getattr(func, '__name__', 'Unknown') : func(y_true = y_true, y_pred = y_pred) for func in self.metrics} Now all we need is a prediction. For this example, I made a quick feature engineering and some hyperparameter tuning. The last step is just to run the benchmark: binary_benchmark = ChurnBinaryBenchmark( metrics=[f1_score, precision_score, recall_score, tp, tn, fp, fn, financial_gain], benchmark_models=[BinaryMean, SimpleXbg, MajorityClass, SimpleKNN, BusinessBenchmark] ) res = binary_benchmark.compare_pred_with_benchmark( df_train=df_train, df_test=df_test, my_predictions=preds, ) pd.DataFrame(res) Benchmark metrics comparison | Image by Author This generates a comparison table of all models across all metrics. Using this table, it is possible to draw concrete conclusions on the model’s predictions and make informed decisions on the following steps of the process. Some drawbacks As we’ve seen there are plenty of reasons why it is useful to have a benchmark. However, even though benchmarks are incredibly useful, there are some pitfalls to watch out for: Non-Informative Benchmark — When the metrics or models are poorly defined the marginal impact of having a benchmark decreases. Always define meaningful baselines. Misinterpretation by Stakeholders — Communication with the client is essential, it is important to state clearly what the metrics are measuring. The best model might not be the best on all the defined metrics. Overfitting to the Benchmark — You might end up trying to create features that are too specific, that might beat the benchmark, but do not generalize well in prediction. Don’t focus on beating the benchmark, but on creating the best solution possible to the problem. Change of Objective — Objectives defined might change, due to miscommunication or changes in plans. Keep your benchmark flexible so it can adapt when needed. Final thoughts Benchmarks provide clarity, ensure improvements are measurable, and create a shared reference point between data scientists and clients. They help avoid the trap of assuming a model is performing well without proof and ensure that every iteration brings real value. They also act as a communication tool, making it easier to explain progress to clients. Instead of just presenting numbers, you can show clear comparisons that highlight improvements. Here you can find a notebook with a full implementation from this blog post. The post How To Build a Benchmark for Your Models appeared first on Towards Data Science.

AI’s thirst for energy is ballooning into a monster of a challenge. And it’s not just about the electricity bills. The environmental fallout is serious, stretching to guzzling precious water resources, creating mountains of electronic waste, and, yes, adding to those greenhouse gas emissions we’re all trying to cut.As AI models get ever more complex and weave themselves into yet more parts of our lives, a massive question mark hangs in the air: can we power this revolution without costing the Earth?The numbers don’t lie: AI’s energy demand is escalating fastThe sheer computing power needed for the smartest AI out there is on an almost unbelievable upward curve – some say it’s doubling roughly every few months. This isn’t a gentle slope; it’s a vertical climb that’s threatening to leave even our most optimistic energy plans in the dust.To give you a sense of scale, AI’s future energy needs could soon gulp down as much electricity as entire countries like Japan or the Netherlands, or even large US states like California. When you hear stats like that, you start to see the potential squeeze AI could put on the power grids we all rely on.2024 saw a record 4.3% surge in global electricity demand, and AI’s expansion was a big reason why, alongside the boom in electric cars and factories working harder. Wind back to 2022, and data centres, AI, and even cryptocurrency mining were already accounting for nearly 2% of all the electricity used worldwide – that’s about 460 terawatt-hours (TWh).Jump to 2024, and data centres on their own use around 415 TWh, which is roughly 1.5% of the global total, and growing at 12% a year. AI’s direct share of that slice is still relatively small – about 20 TWh, or 0.02% of global energy use – but hold onto your hats, because that number is set to rocket upwards.The forecasts? Well, they’re pretty eye-opening. By the end of 2025, AI data centres around the world could demand an extra 10 gigawatts (GW) of power. That’s more than the entire power capacity of a place like Utah.Roll on to 2026, and global data centre electricity use could hit 1,000 TWh – similar to what Japan uses right now. And, by 2027, the global power hunger of AI data centres is tipped to reach 68 GW, which is almost what California had in total power capacity back in 2022. Towards the end of this decade, the figures get even more jaw-dropping. Global data centre electricity consumption is predicted to double to around 945 TWh by 2030, which is just shy of 3% of all the electricity used on the planet.OPEC reckons data centre electricity use could even triple to 1,500 TWh by then. And Goldman Sachs? They’re saying global power demand from data centres could leap by as much as 165% compared to 2023, with those data centres specifically kitted out for AI seeing their demand shoot up by more than four times.There are even suggestions that data centres could be responsible for up to 21% of all global energy demand by 2030 if you count the energy it takes to get AI services to us, the users.When we talk about AI’s energy use, it mainly splits into two big chunks: training the AI, and then actually using it.Training enormous models, like GPT-4, takes a colossal amount of energy. Just to train GPT-3, for example, it’s estimated they used 1,287 megawatt-hours (MWh) of electricity, and GPT-4 is thought to have needed a whopping 50 times more than that. While training is a power hog, it’s the day-to-day running of these trained models that can chew through over 80% of AI’s total energy. It’s reported that asking ChatGPT a single question uses about ten times more energy than a Google search (we’re talking roughly 2.9 Wh versus 0.3 Wh).With everyone jumping on the generative AI bandwagon, the race is on to build ever more powerful – and therefore more energy-guzzling – data centres.So, can we supply energy for AI – and for ourselves?This is the million-dollar question, isn’t it? Can our planet’s energy systems cope with this new demand? We’re already juggling a mix of fossil fuels, nuclear power, and renewables. If we’re going to feed AI’s growing appetite sustainably, we need to ramp up and diversify how we generate energy, and fast.Naturally, renewable energy – solar, wind, hydro, geothermal – is a huge piece of the puzzle. In the US, for instance, renewables are set to go from 23% of power generation in 2024 to 27% by 2026. The tech giants are making some big promises; Microsoft, for example, is planning to buy 10.5 GW of renewable energy between 2026 and 2030 just for its data centres. AI itself could actually help us use renewable energy more efficiently, perhaps cutting energy use by up to 60% in some areas by making energy storage smarter and managing power grids better.But let’s not get carried away. Renewables have their own headaches. The sun doesn’t always shine, and the wind doesn’t always blow, which is a real problem for data centres that need power around the clock, every single day. The batteries we have now to smooth out these bumps are often expensive and take up a lot of room. Plus, plugging massive new renewable projects into our existing power grids can be a slow and complicated business.This is where nuclear power is starting to look more appealing to some, especially as a steady, low-carbon way to power AI’s massive energy needs. It delivers that crucial 24/7 power, which is exactly what data centres crave. There’s a lot of buzz around Small Modular Reactors (SMRs) too, because they’re potentially more flexible and have beefed-up safety features. And it’s not just talk; big names like Microsoft, Amazon, and Google are seriously looking into nuclear options.Matt Garman, who heads up AWS, recently put it plainly to the BBC, calling nuclear a “great solution” for data centres. He said it’s “an excellent source of zero carbon, 24/7 power.” He also stressed that planning for future energy is a massive part of what AWS does.“It’s something we plan many years out,” Garman mentioned. “We invest ahead. I think the world is going to have to build new technologies. I believe nuclear is a big part of that, particularly as we look 10 years out.”Still, nuclear power isn’t a magic wand. Building new reactors takes a notoriously long time, costs a fortune, and involves wading through complex red tape. And let’s be frank, public opinion on nuclear power is still a bit shaky, often because of past accidents, even though modern reactors are much safer.The sheer speed at which AI is developing also creates a bit of a mismatch with how long it takes to get a new nuclear plant up and running. This could mean we end up leaning even more heavily on fossil fuels in the short term, which isn’t great for our green ambitions. Plus, the idea of sticking data centres right next to nuclear plants has got some people worried about what that might do to electricity prices and reliability for everyone else.Not just kilowatts: Wider environmental shadow of AI loomsAI’s impact on the planet goes way beyond just the electricity it uses. Those data centres get hot, and cooling them down uses vast amounts of water. Your average data centre sips about 1.7 litres of water for every kilowatt-hour of energy it burns through.Back in 2022, Google’s data centres reportedly drank their way through about 5 billion gallons of fresh water – that’s a 20% jump from the year before. Some estimates suggest that for every kWh a data centre uses, it might need up to two litres of water just for cooling. Put it another way, global AI infrastructure could soon be chugging six times more water than the entirety of Denmark.And then there’s the ever-growing mountain of electronic waste, or e-waste. Because AI tech – especially specialised hardware like GPUs and TPUs – moves so fast, old kit gets thrown out more often. We could be looking at AI contributing to an e-waste pile-up from data centres hitting five million tons every year by 2030. Even making the AI chips and all the other bits for data centres takes a toll on our natural resources and the environment. It means mining for critical minerals like lithium and cobalt, often using methods that aren’t exactly kind to the planet.Just to make one AI chip can take over 1,400 litres of water and 3,000 kWh of electricity. This hunger for new hardware is also pushing for more semiconductor factories, which, guess what, often leads to more gas-powered energy plants being built.And, of course, we can’t forget the carbon emissions. When AI is powered by electricity generated from burning fossil fuels, it adds to the climate change problem we’re all facing. It’s estimated that training just one big AI model can pump out as much CO2 as hundreds of US homes do in a year.If you look at the environmental reports from the big tech companies, you can see AI’s growing carbon footprint. Microsoft’s yearly emissions, for example, went up by about 40% between 2020 and 2023, mostly because they were building more data centres for AI. Google also reported that its total greenhouse gas emissions have shot up by nearly 50% over the last five years, with the power demands of its AI data centres being a major culprit.Can we innovate our way out?It might sound like all doom and gloom, but a combination of new ideas could help.A big focus is on making AI algorithms themselves more energy-efficient. Researchers are coming up with clever tricks like “model pruning” (stripping out unnecessary bits of an AI model), “quantisation” (using less precise numbers, which saves energy), and “knowledge distillation” (where a smaller, thriftier AI model learns from a big, complex one). Designing smaller, more specialised AI models that do specific jobs with less power is also a priority.Inside data centres, things like “power capping” (putting a lid on how much power hardware can draw) and “dynamic resource allocation” (shifting computing power around based on real-time needs and when renewable energy is plentiful) can make a real difference. Software that’s “AI-aware” can even shift less urgent AI jobs to times when energy is cleaner or demand on the grid is lower. AI can even be used to make the cooling systems in data centres more efficient.On-device AI could also help to reduce power consumption. Instead of sending data off to massive, power-hungry cloud data centres, the AI processing happens right there on your phone or device. This could slash energy use, as the chips designed for this prioritise being efficient over raw power.And we can’t forget about rules and regulations. Governments are starting to wake up to the need to make AI accountable for its energy use and wider environmental impact.Having clear, standard ways to measure and report AI’s footprint is a crucial first step. We also need policies that encourage companies to make hardware that lasts longer and is easier to recycle, to help tackle that e-waste mountain. Things like energy credit trading systems could even give companies a financial reason to choose greener AI tech.It’s worth noting that the United Arab Emirates and the United States shook hands this week on a deal to build the biggest AI campus outside the US in the Gulf. While this shows just how important AI is becoming globally, it also throws a spotlight on why all these energy and environmental concerns need to be front and centre for such huge projects.Finding a sustainable future for AIAI has the power to do some amazing things, but its ferocious appetite for energy is a serious hurdle. The predictions for its future power demands are genuinely startling, potentially matching what whole countries use.If we’re going to meet this demand, we need a smart mix of energy sources. Renewables are fantastic for the long run, but they have their wobbles when it comes to consistent supply and scaling up quickly. Nuclear power – including those newer SMRs – offers a reliable, low-carbon option that’s definitely catching the eye of big tech companies. But we still need to get our heads around the safety, cost, and how long they take to build.And remember, it’s not just about electricity. AI’s broader environmental impact – from the water it drinks to cool data centres, to the growing piles of e-waste from its hardware, and the resources it uses up during manufacturing – is huge. We need to look at the whole picture if we’re serious about lessening AI’s ecological footprint.The good news? There are plenty of promising ideas and innovations bubbling up. Energy-saving AI algorithms, clever power management in data centres, AI-aware software that can manage workloads intelligently, and the shift towards on-device AI all offer ways to cut down on energy use. Plus, the fact that we’re even talking about AI’s environmental impact more means that discussions around policies and rules to push for sustainability are finally happening.Dealing with AI’s energy and environmental challenges needs everyone – researchers, the tech industry, and policymakers – to roll up their sleeves and work together, and fast.If we make energy efficiency a top priority in how AI is developed, invest properly in sustainable energy, manage hardware responsibly from cradle to grave, and put supportive policies in place, we can aim for a future where AI’s incredible potential is unlocked in a way that doesn’t break our planet.The race to lead in AI has to be a race for sustainable AI too.(Photo by Nejc Soklič)Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.Explore other upcoming enterprise technology events and webinars powered by TechForge here.

Ravi Kunju, Chief Product and Strategy Officer, AltairMay 16, 20254 Min ReadZoonar GmbH via Alamy StockCIOs looking to propel digital transformation face a daunting task. They need to harness the power of data analytics and artificial intelligence. They need to empower their employees and serve their customers with the best software tools, compute resources, data infrastructure, and more. Yet to achieve this, CIOs need to navigate a ballooning budget and technology partners who might be hiking prices without giving you more features. A recent Forrester report found that nearly 80% of US organizations reported increases in software costs over the past year. Moreover, 4 in 5 tech leaders said they anticipated their organization’s adoption of generative AI will increase software costs. Prices are going up.  Instead of cutting the technology offered, CIO should explore changing their software licensing to improve their bottom line. Though the board might not get excited about licensing changes, they should. Complex agreements, the wrong licensing models cost an organization dearly -- both financially and technologically.  At the enterprise level, new types of models are emerging that could improve a company’s bottom line, including one that mimics a library system. They’re challenging the traditional per-seat model and could be the future of software licensing.  Related:Traditional Per-Seat Software Licensing for the Right Tools The per seat model, or named user licensing, has its benefits. For instance, when you have a limited set of people who need specific tools, or where there can be individualized attributes or access controls, like an enterprise resource planning system or customer relationship management software. However, there is risk in extending similar models into other domains like engineering simulation. An engineer and their team solving complex problems and running virtual simulations on computers are using several tools, all for different purposes, stages of design, and use cases. The Office of the CFO isn’t using one spreadsheet app -- they’re using CRMs, databases, accounting, and budgeting tools, etc.  And the company’s needs go beyond software. Your organization needs hardware: cloud resources, high-performance computing (HPC) power, mainframes, and beyond.  For the wrong tools, named user licensing can be like owning a car that’s always left running, or it can cause traffic jams for projects if only one person has access to a specific software.  Token-Based Systems Make It Easier to Track Usage Consumption licensing models -- where everyone draws from a shared, but finite “tank” of resources (“fuel,” such as tokens or credits) -- can be an improvement. They house resources under a single licensing framework and allow different user personas to navigate between tools as needed. They also give organizations more insight into costs and usage (since they can see when the tank is getting low). This is a popular cloud licensing model, though the technology is also seeing continued inflation costs.    Related:But the tank for consumption models can run dry very fast. When it’s empty, you have to pay for more fuel. Plus, having that tank isn’t always helpful for controlling costs. What if someone accidentally runs a compute job overnight that uses all your fuel? You can’t get it back -- you have to buy more and hope it doesn’t happen again. There’s never a guarantee you won’t overshoot your budget. Value-Based Models are Interesting, but Subjective Value-based licensing is a relatively newer software licensing concept aligned with business outcomes. The idea is that the cost of the software is based purely on the benefits to the end customer. These models are well-suited for solutions that can easily quantify business benefits; however, these can be subjective and hard to implement and scale. They can also extend the sales cycle and make every transaction unique (and potentially contentious).  Related:A Library Model Combines Other Emergent Licensing Trends Growing in popularity is the flexible-hybrid model, which combines consumption-based and value-based licensing with a predetermined capacity commitment that can’t be exceeded.  It’s like a community e-library, wherein a variety of books can be checked out when needed and checked back in when not in use. In this example, the flexible-hybrid model would work by making certain books available to a team to check out. The only determination then is how many books each team member will borrow at once.  With flexible-hybrid models, CIOs can often get the best of everything: software/hardware resources in a single, shared environment, seamless access to tools/resources when and how you need them, and a brake that prevents runaway costs.  Research from ITAM found that more than 76% of organizations considered themselves “over-licensed.” These emerging software licensing models could fix that number. However, we’ll only see it take off if CIOs start critically evaluating their licensing models and asking their vendors for the flexible-hybrid approach. While AI tops the list of many business leaders’ priorities right now, making sure to secure AI tools that everyone can use, and within budget, should be up there, too.  About the AuthorRavi Kunju Chief Product and Strategy Officer, AltairRavi Kunju is responsible for the strategy and vision of Altair products, which includes facilitating the development, sales, pricing, and marketing of Altair’s solutions for multiple industry verticals, across all business lines. Previously, Kunju held several leadership roles at Altair including vice president of business strategy for Altair’s manufacturing solutions, vice president of strategy for enterprise computing, managing director of enterprise computing for the Americas, and managing director of the U.S. North Central region. Prior to joining Altair, Kunju was a project engineer at Ford Motor Company and a structural analyst at Chrysler Corporation.See more from Ravi KunjuWebinarsMore WebinarsReportsMore ReportsNever Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.SIGN-UPYou May Also Like

Seeking closure Meta argues enshittification isn’t real in bid to toss FTC monopoly trial How many ads is too many? Meta denies ad load harms users in bid to end trial early. Ashley Belanger – May 16, 2025 12:01 pm | 13 Mark Zuckerberg, chief executive officer of Meta Platforms Inc., departs federal court in Washington, DC, US, on Wednesday, April 16, 2025. Credit: Bloomberg / Contributor | Bloomberg Mark Zuckerberg, chief executive officer of Meta Platforms Inc., departs federal court in Washington, DC, US, on Wednesday, April 16, 2025. Credit: Bloomberg / Contributor | Bloomberg Story text Size Small Standard Large Width * Standard Wide Links Standard Orange * Subscribers only   Learn more Meta thinks there's no reason to carry on with its defense after the Federal Trade Commission closed its monopoly case, and the company has moved to end the trial early by claiming that the FTC utterly failed to prove its case. "The FTC has no proof that Meta has monopoly power," Meta's motion for judgment filed Thursday said, "and therefore the court should rule in favor of Meta." According to Meta, the FTC failed to show evidence that "the overall quality of Meta’s apps has declined" or that the company shows too many ads to users. Meta says that's "fatal" to the FTC's case that the company wielded monopoly power to pursue more ad revenue while degrading user experience over time (an Internet trend known as "enshittification"). And on top of allegedly showing no evidence of "ad load, privacy, integrity, and features" degradation on Meta apps, Meta argued there's no precedent for an antitrust claim rooted in this alleged harm. "Meta knows of no case finding monopoly power based solely on a claimed degradation in product quality, and the FTC has cited none," Meta argued. Meta has maintained throughout the trial that its users actually like seeing ads. In the company's recent motion, Meta argued that the FTC provided no insights into what "the right number of ads" should be, "let alone" provide proof that "Meta showed more ads" than it would in a competitive market where users could easily switch services if ad load became overwhelming. Further, Meta argued that the FTC did not show evidence that users sharing friends-and-family content were shown more ads. Meta noted that it "does not profit by showing more ads to users who do not click on them," so it only shows more ads to users who click ads. Meta also insisted that there's "nothing but speculation" showing that Instagram or WhatsApp would have been better off or grown into rivals had Meta not acquired them. The company claimed that without Meta's resources, Instagram may have died off. Meta noted that Instagram co-founder Kevin Systrom testified that his app was “pretty broken and duct-taped” together, making it "vulnerable to spam" before Meta bought it. Rather than enshittification, what Meta did to Instagram could be considered "a consumer-welfare bonanza," Meta argued, while dismissing "smoking gun" emails from Mark Zuckerberg discussing buying Instagram to bury it as "legally irrelevant." Dismissing these as "a few dated emails," Meta argued that "efforts to litigate Mr. Zuckerberg’s state of mind before the acquisition in 2012 are pointless." "What matters is what Meta did," Meta argued, which was pump Instagram with resources that allowed it "to 'thrive'—adding many new features, attracting hundreds of millions and then billions of users, and monetizing with great success." In the case of WhatsApp, Meta argued that nobody thinks WhatsApp had any intention to pivot to social media when the founders testified that their goal was to never add social features, preferring to offer a simple, clean messaging app. And Meta disputed any claim that it feared Google might buy WhatsApp as the basis for creating a Facebook rival, arguing that "the sole Meta witness to (supposedly) learn of Google’s acquisition efforts testified that he did not have that worry." Meta hopes to avoid breakup The monopoly trial is supposed to run through June, but Meta is hoping that US District Judge James Boasberg will agree that the FTC has failed to make its case and end the trial early. Granting Meta's motion would remove any threat of a breakup of its family of apps while also letting Meta off the hook of raising its defense. This could spare Meta any potential further embarrassment of having its founder's unvarnished internal emails picked apart in public. For Boasberg, it will likely come down to the FTC's market definition alleging that Meta dominates and forecloses rivals in a personal social networking services market that Meta says is a "fiction." If Boasberg buys Meta's argument that TikTok is actually Meta's biggest rival—and not Snap or MeWe—Meta could have an easier time shutting down the case. The FTC has not yet commented on the motion, but the agency was determined to go through with the trial, rejecting a reportedly $1 billion settlement offer from Meta. Holding out for $30 billion, the FTC appeared confident in presenting its case, arguing that TikTok—which attracts users broadcasting to strangers—is not a substitute for Meta's apps, which are designed to connect friends and family. The FTC likely expects to see the trial through to the end, when the agency, in a win, will likely try to force Meta to spin off Instagram and WhatsApp. That's not necessarily an inevitability in a Meta loss, though. If Boasberg denies Meta's motion, Meta will have to present more evidence and deliver closing arguments before a second phase of the trial would litigate potential remedies. Ashley Belanger Senior Policy Reporter Ashley Belanger Senior Policy Reporter Ashley is a senior policy reporter for Ars Technica, dedicated to tracking social impacts of emerging policies and new technologies. She is a Chicago-based journalist with 20 years of experience. 13 Comments