TechTarget and Informa Tech’s Digital Business Combine.TechTarget and InformaTechTarget and Informa Tech’s Digital Business Combine.Together, we power an unparalleled network of 220+ online properties covering 10,000+ granular topics, serving an audience of 50+ million professionals with original, objective content from trusted sources. We help you gain critical insights and make more informed decisions across your business priorities.Report: Gamurs Group torches PC Invasion and continues layoffsReport: Gamurs Group torches PC Invasion and continues layoffsThe media conglomerate behind Destructoid and Prima Games has seemingly axed more positions.Diego Arguello, ContributorApril 15, 20253 Min ReadImage via Gamurs GroupGamurs Group, the media conglomerate that owns websites including Destructoid, Prima Games, Twinfinite, and Siliconera, has reportedly laid off more staff across the company.As spotted by GamesIndustry.biz, a number of workers impacted by the latest round of cuts shared the news on Linkedin. That includes Operation Sports assigning editor Christian Smith, Gamurs Group evergreen updates editor Chris Jecks, lead data analyst Thomas Tallentire, SEO manager Jonna Turner, and content strategist Anne Banks.Most of those employees had been working for Gamurs for years. Jecks, for example, started as a contract writer for Twinfinite back in 2015 before rising through the ranks to become contract managing editor in August 2022.Others, however, were let go at the very beginning of their tenure. "Well, I wasn’t anticipating getting this news three months after getting the job, but I've been laid off from my position as Assigning Editor at GAMURS," explained Smith.Gamurs has yet to disclose the number of staff affected or provide a reason for the layoffs—but the company is seemingly well versed in the art of cutting workers loose.In September 2024, Gamurs reportedly laid off 30 people, including staff at Dot Esports, Destructoid, and Twinfinite. As clocked by GamesIndustry.biz, there also appeared to be additional layoffs at Destructoid in February 2025, with more cuts at The Mary Sue and The Escapist following in March.Related:Game media workers have been forced to contend with a growing number of layoffs in recent months. Kotaku Australia was shut down by publisher Pedestrian Group in July 2024, while Ziff Davis conducted soft layoffs in December by asking employees across IGN, Humble, and GamesIndustry.biz to jump ship."PC Invasion is gone"Gamurs' also appears to have completely shuttered its PC-focused publication PC Invasion. Yesterday, ex-employees at the outlet noticed the site is no longer active. It currently redirects to Prima Games."PC Invasion is gone," Digital Trends freelancer and former PC Invasion contributor Anyka Pettigrew wrote on BlueSky. "The gaming website that me and many other writers worked so hard on until GAMURS fired the full-time staff, ghosted the freelancers, and left for dead for months. It just redirects to PrimaGames now. I wrote 1,000+ articles for it btw. It's all gone."Later in the day, Pettigrew noticed that some of the outlet's content archive is progressively appearing on Prima Games. A limited version of the site is also accessible via the Internet Archive.Related:PC Invasion was founded in 2015. At the time of writing, Prima Games hosts author pages of ex-PC Invasion staff members, including founder and editor Paul Younger, and lists articles from 2018 and 2019. The site is also named on Gamurs' brands page."Woke up to find that Gamurs Group has wiped three years of my work from the internet," Insider Gaming senior editor and former PC Invasion hardware editor Mike Straw wrote on BlueSky. "THREE YEARS OF HARD WORK...GONE. [Gamurs Group CEO] Riad Chikhani is nothing more than a snake who has ruined countless outlets. How you sleep at night is beyond me."Game Developer has contacted Gamurs Group for more information.Read more about:LayoffsAbout the AuthorDiego ArguelloContributorSee more from Diego ArguelloDaily news, dev blogs, and stories from Game Developer straight to your inboxStay UpdatedYou May Also Like

Google announced Tuesday that it’s going to redirect country code top-level domain names (ccTLD) for Google Search to google.com over the coming months to “streamline people’s experience on Search,” according to a blog post. The change means that if, for example, you used google.ng (for Nigeria) or google.com.br (for Brazil), you’ll see google.com in your address bar instead. As the change “gradually” rolls out, “you may be prompted to re-enter some of your Search preferences in the process,” the company says. Nothing about your experience should otherwise change. “It’s important to note that while this update will change what people see in their browser address bar, it won’t affect the way Search works, nor will it change how we handle obligations under national laws,” Google says. Google notes that, since 2017, it has provided “the same experience with local results for everyone using Search” if they were using their country’s ccTLD or google.com. “Because of this improvement, country-level domains are no longer necessary.”

Large language models are increasingly used to solve math problems that mimic real-world reasoning tasks. These models are tested for their ability to answer factual queries and how well they can handle multi-step logical processes. Mathematical problem-solving offers a reliable way to examine whether models can extract the necessary information, navigate complex statements, and compute answers correctly. This field has become central to understanding the extent of AI’s logical and cognitive capabilities. A key concern in this domain is how these models perform when their inputs aren’t neat or formatted. In many cases, the questions LLMs encounter in practice come with extra background information, irrelevant details, or even subtle hints that could lead them off track. While models can perform well on standard benchmark problems, their ability to isolate important information from cluttered prompts remains questionable. This has raised the need to examine how distractions influence their reasoning and whether current models are ready for unpredictable, real-world use cases. Past tools and benchmarks have focused mostly on well-formed problem sets, such as GSM8K or MATH. Still, newer variants like GSM-Symbolic and GSM-PLUS began testing model performance under symbolic variations and distractor insertions. These tools uncovered significant weaknesses in LLMs when faced with small changes to the problem text. For instance, introducing one clause that seems relevant but is logically redundant can reduce model accuracy by as much as 65%. This led to the conclusion that models often rely on surface patterns rather than genuine reasoning, which prompted further exploration into more realistic and noisy testing conditions. A team of researchers from the Massachusetts Institute of Technology has introduced a research focused on measuring how LLMs handle four types of systematic perturbations: irrelevant context, pathological instructions, relevant but non-essential information, and a combination of the latter two. The team evaluated 13 large language models—both open-source and commercial—through APIs provided by OpenAI, Anthropic, Cohere, and TogetherAI. Instead of relying on full test sets, the team sampled 56 data points from the GSM8K dataset per experiment, ensuring they captured a balanced distribution of reasoning complexity. To construct these altered prompts, the researchers added dense and irrelevant contexts like Wikipedia pages or financial reports into the input. This took up to 90% of the model’s context window. In the pathological scenario, misleading instructions were appended, designed to manipulate the reasoning path without altering the original question. New details that were factually correct but unnecessary were inserted for the relevant context case to see how the models handled distractions that looked informative. In the final variant, pathological and relevant perturbations were combined, increasing the input complexity while observing how this dual pressure influenced model output. The performance dropped most sharply when irrelevant context was introduced. Across all models, the average accuracy dropped by 55.89%. Pathological instructions caused an 8.52% decline, while relevant context led to a 7.01% decrease. Combining the two types of perturbations produced a 12.91% drop in accuracy. Interestingly, performance didn’t correlate with model size—larger models like Mixtral-8x22B and Command-R-Plus experienced greater regressions compared to some smaller models. Also, the number of reasoning steps in a problem didn’t significantly affect the outcome, suggesting that complexity in logical structure wasn’t the dominant factor in performance variance. This study shows that current large language models, even those with billions of parameters, still struggle when their prompts are altered relatively simply. The researchers from MIT demonstrate that model resilience doesn’t improve significantly with size and that the ability to filter and prioritize information is a major gap in LLM design. These findings push for developing models that are better equipped to deal with cluttered and misleading inputs—an essential step for moving closer to reliable AI in real-world environments. Here is the Paper. Also, don’t forget to follow us on Twitter and join our Telegram Channel and LinkedIn Group. Don’t Forget to join our 90k+ ML SubReddit. NikhilNikhil is an intern consultant at Marktechpost. He is pursuing an integrated dual degree in Materials at the Indian Institute of Technology, Kharagpur. Nikhil is an AI/ML enthusiast who is always researching applications in fields like biomaterials and biomedical science. With a strong background in Material Science, he is exploring new advancements and creating opportunities to contribute.Nikhilhttps://www.marktechpost.com/author/nikhil0980/Reflection Begins in Pre-Training: Essential AI Researchers Demonstrate Early Emergence of Reflective Reasoning in LLMs Using Adversarial DatasetsNikhilhttps://www.marktechpost.com/author/nikhil0980/Transformers Gain Robust Multidimensional Positional Understanding: University of Manchester Researchers Introduce a Unified Lie Algebra Framework for N-Dimensional Rotary Position Embedding (RoPE)Nikhilhttps://www.marktechpost.com/author/nikhil0980/Foundation Models No Longer Need Prompts or Labels: EPFL Researchers Introduce a Joint Inference Framework for Fully Unsupervised Adaptation Using Fine-Tuning and In-Context LearningNikhilhttps://www.marktechpost.com/author/nikhil0980/Reasoning Models Know When They’re Right: NYU Researchers Introduce a Hidden-State Probe That Enables Efficient Self-Verification and Reduces Token Usage by 24%

Traditional RAG vs Graph RAG 1 like April 15, 2025 Share this post Author(s): Kalash Vasaniya Originally published on Towards AI. Why Graph RAG Outperforms Classical Retrieval: A Smarter Path to Context-Rich AnswersSource: From https://x.com/akshay_pachaar If you’re not a member but want to read this article, see this friend link here. Graph RAG is next-level for sure. top-k retrieval in RAG rarely works. Legacy RAG methods depend on selecting the “k” most relevant passages or chunks of text. This has some effectiveness but is soon insufficient if you require a complete, cohesive story. Consider abbreviating a biography where every chapter is dedicated to one accomplishment. If you simply take the most, you will be omitting essential information. This provides you with an incomplete picture and produces answers that may lack vital context or linkages between accomplishments. Source: From https://x.com/akshay_pachaar Graph RAG is not conventional. Rather than directly utilizing the highest k components, it forms an interconnected graph depicting key individuals and how they interconnect based on the source texts. To take an example, if you’re summarizing a life story, Graph RAG builds a complete graph wherein the individual (in the interest of argument, name them P) is connected with all the achievements. The strength of the process is that it can present the complete picture by identifying and maintaining relationships within information that would otherwise be lost. Source: From https://x.com/akshay_pachaar Collecting Entities and Their Relations One of the key steps in Graph RAG is… Read the full blog for free on Medium. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI Towards AI - Medium Share this post

Samsung's newest SSD - the Samsung 990 Evo Plus PCIe 4.0 M.2 NVMe solid state drive - is on sale today. Pick up the 2TB model for $129.99 or, if you can swing it, the 4TB model is also discounted to $259.99. It's currently $40-$70 cheaper than the Samsung 990 Pro and most (if not all) gamers won't notice the difference in performance.Samsung 990 Evo Plus 2TB PS5 SSD for $129.994TB for $259.99Samsung 990 Evo Plus 2TB PCIe Gen 4x4 M.2 SSDSamsung 990 Evo Plus 4TB PCIe Gen 4x4 M.2 SSDThe Samsung 990 Evo Plus is an excellent drive for both your gaming PC and your PlayStation 5 console. It exceeds Sony's minimim speed recommendation for the PS5, boasting sequential speeds of up to 7,250 read and 6,300MB/s write. This is a much faster drive than the 990 Evo non-Pro but not quite as fast as the 990 Pro. The main difference between this drive and the more expensive 990 Pro is that this is a DRAM-less drive. For PS5 performance, it makes no difference. For gaming PCs, the 990 Evo Plus supports HMB (host memory buffer), which makes up for the lack of DRAM by using an inconsequential amount of RAM from your system memory. Gamers will not notice any difference between the two.The Samsung 990 Evo Plus does not have a preinstalled heatsink. However, the 990 Evo Plus SSD is a newer single-sided SSD design that is power efficient and doesn't generate as much heat as SSDs from before. That means you probably don't need to use a heatsink and it should still work perfectly fine in a PS5 console without any thermal throttling. That said, you certainly could for peace of mind and I wouldn't see any disadvantage to that aside from spending an extra $7.More SSDs for PS5Looking for more options? Check out our favorite PS5 SSDs for the PS5 console.Corsair MP600 PRO LPXSee it at AmazonCrucial T500 See it at AmazonWD_Black P40See it at AmazonLexar NM790See it at AmazonWhy Should You Trust IGN's Deals Team?IGN's deals team has a combined 30+ years of experience finding the best discounts in gaming, tech, and just about every other category. We don't try to trick our readers into buying things they don't need at prices that aren't worth buying something at. Our ultimate goal is to surface the best possible deals from brands we trust and our editorial team has personal experience with. You can check out our deals standards here for more information on our process, or keep up with the latest deals we find on IGN's Deals account on Twitter.Eric Song is the IGN commerce manager in charge of finding the best gaming and tech deals every day. When Eric isn't hunting for deals for other people at work, he's hunting for deals for himself during his free time.

This article contains spoilers for the Black Mirror episode “Hotel Reverie.” Black Mirror is about technology but it isn’t about technology, you know what I mean? *puts bong down* Sorry about that, maybe an example from “Hotel Reverie” can better explain. In this season 7 installment, Issa Rae stars as Brandy Friday, a modern day movie star who takes a job as the male lead in a remake of her favorite black-and-white classic film Hotel Reverie. Like many other Black Mirror characters, Brandy is guilty of Not Reading The Instructions and doesn’t realize that “remaking the film” means digitally stepping in to the existing version of it where she must play her part convincingly enough to get to the end. Hotel Reverie isn’t exclusively about technology. The Redream program is merely what sets events into motion where the episode, written by Charlie Brooker and directed by Haolu Wang, can delve into complex questions surrounding digital personhood, copyright law, and love itself. Still, without that tech as a starting point, the story wouldn’t get to explore those questions in the first place. That interplay between sci-fi commentary and the human condition is what drew Issa Rae to the role. “I remember being super excited to get a Black Mirror script and to even be considered for it, and then just immediately being immersed in all the different levels of it. I was gonna say yes anyway, but I was definitely in then,” Rae tells Den of Geek and other outlets at a roundtable junket. A longtime fan of Black Mirror, Rae knew that that the technology Brandy uses to enter into Hotel Reverie looked mighty familiar. A small disk that is placed on one’s temple to engage a virtual reality interface has recurred several times throughout the show’s canon. It first popped up in season 3’s “San Junipero,” then made appearances in season 4’s “USS Callister,” and season 5’s “Striking Vipers” (where it was referred to as an “Experiencer Disk”). A variant of it also turns up in season 7’s “Eulogy.” Manufactured by TCKR Systems, the device has never had an official name until now…sort of. “Charlie [Brooker] calls it a ‘Nubbin,’ but Awkwafina’s character calls it a ‘Mesmerizer.’ I don’t know if that’s the slang version of it,” Rae says. “I didn’t have the ‘real’ version on screen. I don’t know if I’m supposed to reveal this but I had like a fake, smaller version with double-sided tape. So I was very excited to hold a real one more recently.” The “real” Nubbin that Rae got to hold recently was undoubtedly the one used to film this meta advertisement for season 7. @blackmirror issa rae taking a quick trip to another reality #nubbin #blackmirror ♬ original sound – BlackMirror That same ad campaign was accompanied by a Netflix-hosted activation where visitors could experience the device that is definitely called a Nubbin with a brand trademark and everything. Looks like Charlie Brooker got his way on this one. Sorry, Awkwafina’s character! Given that the doohickey we now know as a Nubbin first premiered in San Junipero, its use here has an added bit of resonance. Like that all-time classic episode of the show, Hotel Reverie ends up being a touching love story where two women make a connection outside the bonds of time and space. The story even concludes with a San Junipero easter egg as Brandy Friday’s address is revealed to be “3049 Junipero Drive.” “Charlie said that this is the first episode he wrote for this season. [San Junpiero] is big shoes to fill, so I just hope that people can look at them both separately, but also appreciate them as complements to one another,” Rae says. Join our mailing list Get the best of Den of Geek delivered right to your inbox! All six episodes of Black Mirror season 7 are available to stream on Netflix now.

SummarySatirical sci-fi puzzle game with over 100 handcrafted levels and a satirical A.I. named S.A.R.A.We Interviewed developer Gwen Frey about designing puzzles, building personality through systems, and voicing nearly the entire cast herself.Lab Rat is out Now on Xbox. Gwen Frey is no stranger to big ideas—or big games. After working on titles like BioShock Infinite and The Flame in the Flood, she’s back with a new game! Lab Rat, a satirical Sci-Fi puzzle game launching on Xbox that pits players against an overly confident AI named S.A.R.A. With over 100 handcrafted puzzles, sharp writing, and a surprising amount of heart, Lab Rat is as much about logic as it is about being observed. I sat down with Gwen to talk puzzle design, working solo (and not-so-solo), and what it’s like to voice nearly every character in your own game. Corey: S.A.R.A. plays a central role in Lab Rat—a hyper-intelligent, metrics-obsessed AI who observes and reacts to everything the player does. Inspired in part by your experience working alone during lockdown, she brings humor, tension, and a sense of being constantly evaluated. Did the concept for S.A.R.A. shape the gameplay, or did the puzzles come first? Gwen: The idea for S.A.R.A. developed naturally alongside the game’s core mechanics. Early on, my puzzle designer and I were exploring a new system, but thematically we both felt like rats in a cage—trapped inside, isolated. We built that feeling into everything: puzzles that evoke confinement, jittery character animations, and even blocks that electrocute you when misused, like a science experiment gone wrong. Visually, I was inspired by my own dual-monitor setup during lockdown, when most of my interaction was through screens. That evolved into the concept of S.A.R.A.—an overbearing algorithm watching your every move, built to reflect the eerie, disconnected mood we were living through. Once I saw the game world through that lens, everything started to click. Corey: You’ve described Lab Rat as a game that “makes people think and laugh.” Which of those two was harder to design for? Gwen: Definitely making people laugh! When a person sits down to complete a puzzle game they are prepared to think deeply and solve problems – they are in the correct mindset for that. Thought-provoking commentary is generally a welcome addition to that experience. However, comedy is unexpected, very difficult to do well, and different kinds of humor appeal to different people. Seriously – comedy is hard. Corey: There’s a really clever rhythm to how puzzles unfold in Lab Rat. What was your approach to pacing the difficulty curve? Gwen: There are two ways I like to keep puzzle players engaged: discovery and mastery. Discovery is when you encounter something new and figure out how it works; mastery is using that understanding to solve a tougher challenge. I try to introduce something new every few puzzles to keep things fresh, and I aim for solutions that feel earned—clear enough to grasp with some thought, but never obvious. Block-pushing games come with a unique problem: it’s easy to end up in an unwinnable state without realizing it. That kind of thing kills motivation. So in Lab Rat, we let you undo moves, pull blocks off walls—basically give you room to experiment without fear of getting stuck. It makes the game more accessible, especially for players new to the genre, without making it any less satisfying. Corey: Working on a game in a small team means wearing a lot of hats—designer, writer, programmer. Which part of that process felt most natural to you, and which one surprised you the most? Gwen: Programming, art, and animation all came very naturally to me. These are things I’ve done for years, and I’m working in the Unreal engine, which I’ve used my entire career, so there was nothing new to learn. By far the most surprising thing I had to learn for this project was voice acting. I am not a performer and I have never done voice work before Lab Rat. It was quite challenging. Corey:  We’ve been fortunate enough to have played through the game, so we’ve seen the end credits.  Exactly how many characters did you personally voice in the game? Have you counted? Gwen: This question made me laugh. My programmer put my name in the credits about 100 times as a sort of joke. I might change that before launch… but it is true that I voiced every character in Lab Rat. There was one line where we absolutely needed a male voice and my husband stepped up to the mic for it, but other than that all the characters were voiced by me personally. I spent a lot of time recording arguments with myself, and then made extensive use of sound effects in the engine to make each character sound unique. Corey: If you could sit next to someone playing Lab Rat on Xbox for the first time— without spoiling it, is there a particular moment you would be waiting for them to reach? Gwen: There are many, many moments I hope they reach! I wanted every 30 minutes of Lab Rat to feel more interesting and unexpected than the previous 30 minutes. I’m excited to see if we achieved this. Corey: Huge thanks to you, Gwen, for taking the time to chat about Lab Rat, indie game development, and what it’s like to argue with yourself in a recording booth! Lab Rat is available now on Xbox One optimized for Xbox Series X|S. Lab Rat Klei Publishing ☆☆☆☆☆ ★★★★★ $19.99 Get it now Congratulations! You have been chosen to participate in a special test that will help refine a brand new kind of game! Lab Rat was generated by the world's most advanced machine learning algorithm – me. I have been meticulously trained on the best interactive entertainment available today and my data indicates you will be completely satisfied with the result. However, your valuable human feedback is required to help me further adjust and develop this experience. Key Features: Maneuver objects, divert lasers, and manipulate electricity to overcome over a hundred surprising genre-bending puzzles. Outwit a metrics-obsessed machine intent on heckling you for your humanity. Participate in satirical in-game analytics that are updated in real time with real player data. The Game: Lab Rat is a hand-crafted narrative puzzler masquerading as a machine-generated video game. This satirical adventure stars a metrics-obsessed AI who will monitor, profile, and guide you as you solve over a hundred unique spatial problems. Over time this AI will develop a hilariously-misguided understanding of humanity based on your performance and survey response data. The Team: While Lab Rat pokes fun at the absurdity of algorithmically-generated interactions, this game is lovingly hand-crafted by a team of expert human developers. Lab Rat is directed by Gwen Frey (creator of Kine & co-creator of The Flame in the Flood). It features environment art by Mike Snight (Lead World Builder, Bioshock Infinite), writing and programming by Ian Bond (Programmer/Developer, Bioshock 1), and puzzles by Lucas Le Slo (beloved experimental puzzle designer).

Today Apple updated its list of vintage and obsolete products, and the latest additions include the end of an era for the Intel Mac mini. iPhone 6s and the last Mac mini with Intel are now ‘vintage’ Apple periodically adds devices to its ever-growing list of vintage and obsolete products. That happened today, as spotted by MacRumors, with two noteworthy “vintage” additions: iPhone 6s Mac mini (2018) The latter product is especially significant, because the 2018 Mac mini was the last remaining Intel model that was not yet labeled either vintage or obsolete. Apple applies these designations based on the amount of time that’s passed since a product was last available for sale. So what are those timelines exactly? Per Apple’s definitions: Vintage: “Apple stopped distributing them for sale more than 5 and less than 7 years ago.” Obsolete: “Apple stopped distributing them for sale more than 7 years ago.” Availability of service and parts is tied to vintage and obsolete status Even though the iPhone 6s launched back in 2015, it’s joining the list at the same time as the 2018 Mac mini because Apple continued selling that iPhone much longer than its final Intel Mac mini. Since these products are now considered vintage, Apple no longer guarantees that parts for repairs will be readily available. For the first 5 years from a product’s last distribution, Apple does promise to make service and parts available. But now that these devices are considered vintage, you may or may not be able to get support. After the 7 year “obsolete” threshold kicks in, Apple will officially no longer offer service or parts. Currently out of Intel Mac minis, the 2018 and 2014 models are vintage, while all older models are obsolete. Are you still hanging on to an Intel Mac mini? Let us know in the comments. Best Mac accessories Add 9to5Mac to your Google News feed.  FTC: We use income earning auto affiliate links. More.You’re reading 9to5Mac — experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out our homepage for all the latest news, and follow 9to5Mac on Twitter, Facebook, and LinkedIn to stay in the loop. Don’t know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

As times change, so do mission statements, especially in the fast-and-loose world of tech. In recent months, we've seen Google walk back its pledge to "do no evil," and OpenAI quietly delete a policy prohibiting its software's use for "military technology."Mark Zuckerberg's Facebook is no exception. Its 2008 motto, "Facebook helps you connect and share with the people in your life," is now a distant memory — according to Zuckerberg himself, who testified this week that Facebook's main purpose "wasn't really to connect with friends anymore.""The friend part has gone down quite a bit," Zuckerberg said, according to Business Insider.Instead, he says that the platform has evolved away from that model — its original claim to fame, as old heads will recall — in its over 20 years of life, becoming "more of a broad discovery and entertainment space," which is apparently exec-speak for "endless feed of AI slop."The tech bigwig was speaking as a witness at a federal antitrust case launched by the Federal Trade Commission against Meta, the now-parent company to WhatsApp, Instagram, Threads, and Oculus.The FTC's case hinges on a series of messages sent by Zuckerberg and his executives regarding a strategy of buying other social media platforms outright, rather than compete with them in the free and open market — a scheme that's more the rule than the exception for Silicon Valley whales like Google, Amazon, and Microsoft.The FTC alleges that Meta began its monopolistic streak as early as 2008, when Zuckerberg buzzed that "it's better to buy than compete" in a series of emails about then-rival platform Instagram. He finally got its hands on Instagram in 2012, after sending a memo that Facebook — which changed its name to Meta in 2021 — "had" to buy the photo-sharing app for $1 billion, fearing competition and a bidding war with fast-growing platforms like Twitter."The businesses are nascent but the networks are established," Zuckerberg wrote in a leaked email about startup platforms Instagram and Path. "The brands are already meaningful and if they grow to a large scale they could be very disruptive to us.""It’s an email written by someone who recognized Instagram as a threat and was forced to sacrifice a billion dollars because Meta could not meet that threat through competition,” said the FTC’s lead counselor, Daniel Matheson.Those internal memos are now smoking guns in what could be the biggest antitrust case since the infamous AT&T breakup of 1982, which had many similarities to the FTC's suit against Meta. Back then, AT&T held unrivaled market influence that it used to box out smaller fish and shape laws to its whims — to chase profit above all, in other words.Meta, in parallel, has spent millions lobbying lawmakers, is the dominant player in online advertising, and currently wields a market cap of $1.34 trillion — higher than the value of all publicly traded companies in South Korea, for perspective.The FTC's challenge will depend on whether federal prosecutors can convince US District Judge James Boasberg that Meta's acquisitions of Instagram and WhatsApp were illegal by notoriously weak US antitrust standards. They'll have no help from Boasberg, an Obama appointee, who has voiced skepticism with cases against Meta in the past."The [FTC] faces hard questions about whether its claims can hold up in the crucible of trial," Boasberg said in late 2024, adding that "its positions at times strain this country’s creaking antitrust precedents to their limits."Whatever happens, it's clear that Zuckerberg has moved on from the idealism of the early internet — to the sloppified money-grubbing of whatever it is we have now.Share This Article

Everybody knows browser extensions are embedded into nearly every user's daily workflow, from spell checkers to GenAI tools. What most IT and security people don't know is that browser extensions' excessive permissions are a growing risk to organizations. LayerX today announced the release of the Enterprise Browser Extension Security Report 2025, This report is the first and only report to merge public extension marketplace statistics with real-world enterprise usage telemetry. By doing so, it sheds light on one of the most underestimated threat surfaces in modern cybersecurity: browser extensions. The report reveals several findings that IT and security leaders will find interesting, as they build their plans for H2 2025. This includes information and analysis on how many extensions have risky permissions, which kinds of permissions are given, if extension developers are to be trusted, and more. Below, we bring key statistics from the report. Highlights from the Enterprise Browser Extension Security Report 2025 1. Browser extensions are ubiquitous in enterprise environments. 99%, nearly all, of employees, have browser extensions installed. 52% have more than 10 extensions installed. Security analysis: Nearly all employees are exposed to browser extension risk. 2. Most extensions can access critical data. 53% of enterprise users' extensions can access sensitive data like cookies, passwords, web page contents, browsing information, and more. Security analysis: An employee-level compromise could jeopardize the entire organization. 3. Who publishes these extensions? Who knows? More than half (54%) of extension publishers are unknown and only identified via Gmail. 79% of publishers only published one extension. Security analysis: Tracking the reputability of extensions is difficult, if possible at all with IT resources. 4. GenAI extensions are a growing threat. Over 20% of users have at least one GenAI extension, and 58% of these have high-risk permission scopes. Security analysis: Enterprises should define clear policies for GenAI extension use and data sharing. 5. Unmaintained and unknown browser extensions are a growing concern. 51% of extensions haven't been updated in over a year, and 26% of enterprise extensions are sideloaded, bypassing even basic store vetting. Security analysis: Extensions can be vulnerable even if they're not purposefully malicious. 5 Recommendations for Security and IT The report not only brings data, it also provides actionable guidance for security and IT teams, recommending how to deal with the browser extension threat. Here's what LayerX advises organizations: Audit all extensions - A full picture of extensions is the foundation for understanding the threat surface. Therefore, the first step in securing against malicious browser extensions is to audit all extensions in use by employees. Categorize extensions - Certain types of extensions that make them appealing to attack. This can be due to their broad user base (such as GenAI extensions) or because of the permissions granted to such extensions. Categorizing extensions can help assess the browser extension security posture. Enumerate extension permissions - The next step is to list the information extensions can access. This helps further map the attack surface and configure policies later on. Assess extension risk - Now it's time for risk management. This means assessing the risk for each extension based on their permissions and the information they can access. In addition, a holistic risk assessment includes external parameters such as reputation, popularity, publisher, and installation method. Together, these parameters should be combined into a unified risk score. Apply adaptive, risk-based enforcement - Finally, organizations can use their analysis to apply adaptive, risk-based enforcement policies tailored to their uses, needs, and risk profile. Access the Report Browser extensions are not just a productivity tool, they're an attack vector most organizations do not know exists. LayerX's 2025 report provides comprehensive findings and data-driven analysis to help CISOs and security teams rein in this risk and build defensible browser environments. Download the full report. Found this article interesting? This article is a contributed piece from one of our valued partners. Follow us on Twitter  and LinkedIn to read more exclusive content we post.