At Amazons annual Devices & Services event on Wednesday, the company introduced Alexa+, an enhanced version of its voice assistant, now powered by generative AI.During the demonstration, Amazon showcased how users can share documents with Alexa+, allowing it to recall important details and answer questions about those documents.Mara Segal, director of Alexa, provided several examples of how this feature works. In one instance, she asked Alexa+, From grandmas zucchini bread recipe, how much oil did it need? Alexa+ was able to extract the answer from the recipe that had been previously uploaded.In a more complex scenario, a user can upload a document from their Homeowners Association (HOA) and ask questions about the guidelines, which many people tend to overlook.Additionally, users can forward multiple emails from a childs school to Alexa+, extracting and summarizing the essential information. It can also help manage their calendars to ensure they dont miss important school events.Amazon demonstrated several Alexa+ features at the event, including the ability to jump to different movie scenes on Prime Video and control smart home devices, allowing users to move music between speakers in different rooms.

Amazon is not giving up on making its Alexa a virtual shopping assistant, On Wednesday, the company announced that Alexa+, its revamped Alexa experience, will be able to help consumers grocery shop from home using more natural conversations and requests.The feature will work with Amazon Fresh or any other grocery partner working with Amazon, the retailer noted. Food delivery from partners like Uber Eats is also available. Instead of simply listing out ingredients you need by voice, Amazon showed how you could also chat with Alexa as she helps you build your shopping list. For example, the company demoed asking Alexa to get everything we need for banana bread, except the spices, and said they also wanted to cook that Bobs Red Mill pancake mix. Oh, and they needed syrup and a few more eggs, too.Alexa understood these natural language requests and turned them into grocery list items.Plus, Alexa was able to adjust quantities on the fly when asked to make it two and a half gallons instead of a gallon of milk, for instance. The assistant was also able to make recommendations of what to cook for a five-year-old dinner guest by suggesting some pasta options that kids tended to like. The company described this type of shopping as using your voice to shop in a sort of stream of consciousness-style chat. The upgraded Alexa+ experience will ship in March as an upgrade to existing Alexa devices. The service will be $19.99 per month but is free for all Prime members.

Breakdown & ShowreelsPaddington in Peru: VFX Breakdown by FramestoreBy Vincent Frei - 26/02/2025 How do you bring Paddingtons heartwarming adventure to life? This VFX breakdown by Framestore takes you deep into the magic of Paddington in Peru, showcasing breathtaking environments, the dynamic boat sequence, and the attention to detail that makes our favorite bear so real.WANT TO KNOW MORE?Framestore: Dedicated page about Paddington in Peru on Framestore website. Vincent Frei The Art of VFX 2025

Researchers at Purdue Applied Research Institute (PARI) are exploring methods to 3D print dark ceramics, enabling complex, heat-resistant components for hypersonic vehicles.Their objective is to develop complex shapes for hypersonic vehicle components while improving scalability and efficiency. Led by Professor Rodney Trice from Purdues School of Materials Engineering, the team is using Digital Light Processing (DLP) to adapt these ceramics for additive manufacturing. This 3D printer is housed at PARIs Hypersonics Advanced Manufacturing Technology Center (HAMTC).Because dark powders absorb the UV light that would be necessary to cure the material, we cannot form as thick of a layer, said Trice. Therefore, we get cure depths that are too thin, which then negatively impacts the time it takes to build each part.Matthew Thompson loading a crucible into a box furnace to heat and removing binders from 3D printed ceramic samples. Photo via PARI.Overcoming challenges in 3D printing dark ceramicsAccording to Trice, this approach enables the creation of intricate designs with exceptionally smooth surfaces and micron-level precision. His team has successfully printed a variety of shapes, including sharp cones and hemispheres, both crucial components in hypersonic vehicle design.But the process isnt without challenges. Unlike lighter-colored ceramics, which reflect and scatter UV light to cure evenly, dark ceramics absorb light, making it harder to solidify each layer properly. To tackle this, Trice, along with doctoral candidate Matthew Thompson and ceramics research engineer Dylan Crump, is experimenting with different resin formulations and surface treatments to enhance the curing process.Thompson has described the teams work as an ongoing effort to refine the material properties and printing process. He explained that the researchers have been modifying the ceramics to enhance their printability while also addressing challenges in post processing. These include delamination and cracking, issues that become more pronounced with larger components.As the project progresses, scaling up production remains a key focus. Moving from small-scale to large-scale manufacturing presents additional risks, and the team is working to mitigate these challenges.Thompson has emphasized that their aim is to either set up a pipeline to make these parts or find strategies that actual stakeholders can use. This would further help in reducing the time and effort required for research and development (R&D) in similar applications.As per the team, this research is part of a broader initiative supported by the Office of the Secretary of Defense Manufacturing Science and Technology Program. It is being conducted in alliance with the Naval Surface Warfare Center, Crane Division, and the National Security Technology Accelerators Strategic and Spectrum Missions Advanced Resilient Trusted Systems.The findings are expected to contribute to the broader adoption of advanced ceramics in hypersonic vehicle development, bridging the gap between research and real-world application.High-performance ceramics for hypersonic vehiclesPrevious news highlighted the use of ceramics as ideal for hypersonic applications due to their ability to endure extreme heat and pressure while supporting the scalable production of complex, high-precision components.Few years back, the U.S. Air Force (USAF) was exploring 3D printed ceramics for hypersonic flight, by working with HRL Laboratories under a Collaborative Research and Development Material Transfer Agreement (CRADA-MTA). This research focused on silicon oxycarbide (SiOC), a ceramic capable of withstanding extreme temperatures up to 3,200F while enabling complex geometries through 3D printing.An artists concept of the XS-1 spaceplane. Image via DARPA. Scientists at the Aerospace Systems Directorate (ASD) tested SiOC components, including thermocouple radiation shields, to evaluate their structural integrity under hypersonic conditions. For this research, the material was initially printed as a pre-ceramic resin and then heat-treated to form a fully ceramic structure. The CRADA-MTA enabled the Air Force to provide research insights while HRL retained patent rights, supporting further exploration of hypersonic technology.In 2016, Imperial College London (ICL) researchers identified tantalum carbide and hafnium carbide as refractory ceramics with the highest recorded melting points, reaching 4000C. This discovery had the potential to enhance thermal protection structures for hypersonic vehicles, which experience extreme heat exceeding 2200K at their nose tips and leading edges due to velocities above Mach 5.Known for their heat resistance, these ceramics complement advanced manufacturing techniques like 3D printing and ceramic matrix composites used in jet engines, improving efficiency and durability in high-temperature environments. The study was published in Scientific Reports.What3D printing trendsshould you watch out for in 2025?How is thefuture of 3D printingshaping up?To stay up to date with the latest 3D printing news, dont forget to subscribe to the 3D Printing Industry newsletter or follow us on Twitter, or like our page on Facebook.While youre here, why not subscribe to our Youtube channel? Featuring discussion, debriefs, video shorts, and webinar replays.Featured image shows Matthew Thompson loading a crucible into a box furnace to heat and removing binders from 3D printed ceramic samples. Photo via PARI.

Old Place, New Faceadamo-faiden and Chamber Projects bring a taste of Italy to the Argentine capitalByDaniel Jonas Roche February 26, 2025Interiors, International (Javier Agustin Rojas)SHAREOn the corner of Guatemala and Gurruchaga Streets in Buenos Aires, chefs in white jackets lift Neapolitan pizza pies with heavy peels into a flaming hot oven as fashionable patrons observe the starry night sky through textural metal mesh high above their heads. The design of Orno Palermo byadamo-faidenandChamber Projectsoffers an elevated dining experience in Palermo, a historic neighborhood in the Argentine capital. There, the local design offices inserted a new 2-story tapering glass volume behind the load-bearing facade of a quirky 19th-century building. The new location marks Ornos third outpost in Buenos Aires. (The other two are farther north in Villa Urquiza and Belgrano.) From the exterior, the restaurant is shaped like a frustuma pyramid with its apex chopped offand the oven occupies a central location underneath it. Should the space ever be used for religious purposes, the oven is where the priest, imam, rabbi, or pujari would stand; the wooden chairs and tables, the pews. An aluminum chimney connects both floors and punctures the frustum, a compelling architectural feature that makes the overall ensemble recall Le Corbusiers chapel in Firminy, albeit conceptually. A bespoke stair connects the first and second floors. Wood furniture throughout is complemented by white tiles, stainless-steel appliances, and exposed concrete.Read more about on the restaurant on aninteriormag.com. Buenos AiresRestaurants

close From CAPTCHA to catastrophe: How fake verification pages are spreading malware Hackers are using CAPTCHAs to infect your PC with malware. We saw a rise in infostealer malware in 2024, with hackers using it to steal credentials, cryptocurrency and other personal data from millions of users. If you recall, I reported countlessincidents of an infostealer called Lumma preying on Android, Windows and even iOS and Mac users.A new cybersecurity report now highlights that hackers using Lumma, along with StealC, Redline and other infostealers, infected 4.3 million machines in 2024, leading to an astonishing 330 million compromised credentials.Security researchers have also observed 3.9 billion credentials shared in credential lists that appear to be sourced from infostealer logs.STAY PROTECTED & INFORMED! GET SECURITY ALERTS & EXPERT TECH TIPS SIGN UP FOR KURTS THE CYBERGUY REPORT NOW Illustration of a hacker at work (Kurt "CyberGuy" Knutsson)Infostealer-related attacks on the rise in 2024Acybersecurity report by threat intelligence platform KELA has uncovered a sharp rise in infostealer malware in 2024. Researchers also observed an alarming trend in how stolen data was circulated. Large compilations of credentials, often referred to as "credential lists," were being shared across cybercrime forums. These lists, primarily sourced from infostealer logs, contained billions of login details harvested from infected devices.One of the most notable incidents linked to infostealer malware was the breach of Snowflake, a cloud data storage provider. In April 2024, threat actors gained access to customer accounts using stolen login credentials, many of which were obtained through infostealers. Exploiting weak security practices, such as the absence of multifactor authentication, attackers extracted valuable data and later attempted to sell it on underground markets. The breach affected at least 165 companies.WHAT IS ARTIFICIAL INTELLIGENCE (AI)?The KELA report highlights that hackers deploying Lumma, StealC, Redline and other infostealers infected 4.3 million machines, leading to the compromise of 330 million credentials. Nearly 40% of these infected machines contained credentials for corporate systems, including content management platforms, email accounts, Active Directory Federation Services and remote desktop environments. In total, this accounted for 1.7 million compromised bots and 7.5 million stolen credentials.The report also found that 3.9 billion credentials were shared in credential lists that appear to be sourced from infostealer logs. KELAs analysis suggests that almost 65% of infected devices were personal computers storing corporate credentials, making them a prime target for infostealer malware. Illustration of a hacker at work (Kurt "CyberGuy" Knutsson)2025 is not going to be any differentInfostealer malware is not going anywhere in 2025. With malware-as-a-service platforms on the rise and infostealers becoming more advanced, cybercriminals will likely keep using them as a go-to method for stealing credentials and gaining access to systems.Law enforcement has been cracking down, though. In 2024, authorities managed to take down key parts of the infostealer ecosystem, including disrupting Redline, one of the most widely used infostealers. This showed that international agencies can go after not just the malware developers but also the networks and underground markets that keep these operations running.But takedowns like these rarely put an end to the problem. When one major infostealer operation is shut down, others quickly step in to take its place. The constant demand for stolen credentials and the ability of cybercriminals to adapt means infostealer attacks will likely remain a major threat in 2025.GET FOX BUSINESS ON THE GO BY CLICKING HERE Illustration of a hacker at work (Kurt "CyberGuy" Knutsson)Ways to stay safe from infostealer malwareWith infostealer malware becoming a growing threat, protecting your data requires a mix of smart security habits and reliable tools. Here are some effective ways to keep your information safe.1. Enable two-factor authentication (2FA):Even if your credentials are stolen,2FA adds an extra layer of security by requiring a second form of verification, such as a code from an authentication app or biometric confirmation. Cybercriminals rely on stolen usernames and passwords to break into accounts, but with 2FA enabled, they cannot gain access without the additional security step. Make sure to enable 2FA on important accounts like email, banking and work-related logins.2. Use strong antivirus software and be cautious with downloads and links: Infostealer malware often spreads through malicious downloads, phishing emails and fake websites. Avoid downloading software or files from untrusted sources and always double-check links before clicking them. Attackers disguise malware as legitimate software, game cheats or cracked applications, so it is best to stick to official websites and app stores for downloads.The best way to safeguard yourself from malicious links that install malware, potentially accessing your private information, is to have strong antivirus software installed on all your devices. This protection can also alert you to phishing emails and ransomware scams, keeping your personal information and digital assets safe. Get my picks for the best 2025 antivirus protection winners for your Windows, Mac, Android and iOS devices.3. Use a password manager: Many infostealers target saved passwords in web browsers. Instead of relying on your browser to store credentials, use a dedicated password manager. Get more details about mybest expert-reviewed password managers of 2025 here.4. Keep software updated:Cybercriminals exploit outdated software to deliver malware.Keeping your operating system, browsers and security software up to date ensures that known vulnerabilities are patched. Enable automatic updates whenever possible and install reputable antivirus or endpoint protection software that can detect and block infostealer threats before they compromise your system.Kurts key takeawayGiven the surge in infostealer malware warnings, it is clear that cybercriminals are actively targeting passwords. Both organizations and individuals are urged to strengthen their security measures by enabling 2FA, monitoring credential exposure and using endpoint protection tools. While no security measure is completely foolproof, combining these practices can significantly reduce the risk of falling victim to infostealer malware.Do you feel that companies are doing enough to protect your data from infostealer malware and other cyber threats? Let us know by writing us atCyberguy.com/Contact.For more of my tech tips and security alerts, subscribe to my free CyberGuy Report Newsletter by heading toCyberguy.com/Newsletter.Ask Kurt a question or let us know what stories you'd like us to cover.Follow Kurt on his social channels:Answers to the most asked CyberGuy questions:New from Kurt:Copyright 2025 CyberGuy.com.All rights reserved. Kurt "CyberGuy" Knutsson is an award-winning tech journalist who has a deep love of technology, gear and gadgets that make life better with his contributions for Fox News & FOX Business beginning mornings on "FOX & Friends." Got a tech question? Get Kurts free CyberGuy Newsletter, share your voice, a story idea or comment at CyberGuy.com.

When VMware launched vSAN Max in late 2023, it promised storage for VSphere with many of the benefits of hyper-converged infrastructure (HCI), but without some of its more common limitations.With vSAN Max, VMware aims to provide more flexibility over storage deployment for its virtual machine environments, by disaggregating the storage layer.Here, we look at some of the pros and cons of vSAN Max, as well as commercial considerations chief information officers (CIOs) need to be aware of following VMwares acquisition by Broadcom.VMwares vSAN Max aims to square a circle, or perhaps several circles, when it comes to storage for virtual machines. The idea is to create a disaggregated, storage-only cluster for vSphere clusters, which can scale up to petabytes.It sets out to preserve many of the advantages of HCI and the firms vSAN storage technology, but without some of the rigidity that comes with HCI.VMwares vSAN offered a number of benefits. Chief among these were simplicity and ease of deployment. It builds on vSANs Express Storage Architecture and the earlier HCI Mesh.As GigaOm lead analyst Howard Holton points out, before vSAN, most storage systems were not optimised for VMs. They needed careful configuration to work well. Nor was VMware a tier 1 customer for the storage vendors. HCI changed all that, he says. It was an out-of-the-box, VM-aligned and optimised solution that gave companies the easy button. The big issue we were left with was having to buy storage and compute together, regardless of needs.Read more about VMware and storageVMware backup: Key decision points if you migrate away from VMware. VMwares pricing changes have spurred some organisations to move to new virtualisation environments, but theyll need backing up. We look at the key points to consider.VMware vSAN vs. SAN: What are the differences? VMware vSAN differs from a traditional SAN in a few crucial ways, including pooled vs. isolated storage, access management and optimisation for different types of environments.With enterprises storing ever larger volumes of data, this is no longer sustainable. So, with vSAN Max, VMware moved away from the purists idea of HCI and split storage from compute.By disaggregating storage from compute, or resorting to a traditional split of compute and storage needs, you enable a better flexibility in design, says Tom Howarth, principal consultant at Sjultra. Compute nodes can be scaled to needs, and the storage can also be scaled to needs.But critically vSAN and vSAN Max can still be controlled from the same VMware management console.VMware designed vSAN Max to provide a more flexible storage option to VM users, so it is not targeted at any one application.The most obvious use case for vSAN Max is where firms outgrow the built-in storage capacity of their existing HCI and vSAN systems. VMware customers can now scale storage independently of compute. Or, if their compute requirements rise, they scale compute without paying for redundant storage.But vSAN Max should also allow a more granular approach to storage design, optimising performance and cost. With disaggregated storage, you can optimise to the requirements of the workload, such as faster disk (NVMe) for database workloads, slower (SSD) for file and print, and spinning rust (HHD and SSD cache) for archival and backup purposes, says Howarth.Unsurprisingly, the main application for vSAN Max is storage for VMs and it is optimised for this but it can also act as an object file store or even basic file and print.And, as GigaOms Holton points out, vSAN Max also supports higher-end storage features that include deduplication, encryption, compression and erasure coding.Deploying vSAN Max is fairly straightforward. Users create a cluster of new nodes in VMwares admin system, vCenter, and select vSAN Max. They then have the option of local deployments or a stretched cluster. The former needs four or seven nodes depending on functions. Stretched clusters need at least eight nodes across two sites.Users can deploy vSAN Max on commodity hardware, but consultants recommend certified vSAN-ready nodes to guarantee performance.In the enterprise, vSAN Max faces competition from a range of options, primarily those that support vSphere ESXi hosts. These include arrays from hardware suppliers including HPE and Pure Storage.Systems running on commodity hardware also compete with vSAN and vSAN Max. These include software-defined storage suppliers such as Nutanix, StarWind and Nexenta. HCI products from Scale Computing and HPE also compete.As this is disaggregated storage, theres the traditional SAN and NAS providers, like Dell, NetApp, IBM and Pure Storage, says Howarth.Even non-vSAN ready storage can connect via iSCSI to a vSAN Max cluster, he notes.The hardware options vSAN Max offers will certainly benefit enterprises that want to optimise their VM storage, especially those who have already invested in vSAN.But Broadcoms takeover of VMware and the resultant uncertainty over product lines and licence costs is likely to prompt CIOs to look again, and perhaps consider more generic and therefore more flexible options.The biggest issue to me is the lack of trust in Broadcom and VMware, says GigaOms Holton. Broadcom has eroded the confidence and trust customers and partners had in VMware. Increasing your investment in VMware may not be the smart move today.

Apple left MagSafe out of its latest budget handset, but you can still add it with these simple solutions.

The clock is ticking for Kindle users. After February 2025, a long-standing feature disappears. Here's why it matters.

Modern server room, corridor in data centre with Supercomputer racks, neon lights and conditioners. ... [+] 3D rendering illustrationgettyAround the world, the biggest players in LLM technology are coming out with new versions of their models at staggering speed.But how do they stack up?Analyst and testers (and others) are coming up with brand-new evaluations of these competing models and detailing their performance on everything from deep PhD.-level questions, to coding, to various types of specialized tasks.But in the end, some claim that most of this hard work doesnt really make any difference to the average end user. Lets explore this a little bit through the lens of one of my favorite podcasts.Grok-3 and o3: AI Daily Brief ObservesTwo of the standouts right now are OpenAIs o3 mini model and Grok3, the new version of the Xai chatbot that has its own reasoning capabilities and new functionality built in.We can see graphs of these models using GPQA, a Graduate-Level Google-Proof benchmark, and the American Invitational Mathematics Examination (AIME) data set dating back to 1983. Some team members at OpenAI claim that o3 mini is better across the board others at XAI, unsurprisingly, disagree.And then theres the third argumentAI Daily Brief CoverageOver at the AI Daily Brief podcast, host Nathaniel Whittemore covers these types of evolution, starting with a quote by Matthew Lambert:Frankly, there are no industry norms to lean on. Just expect noise. It's fine. May the best models win. Do your own evals anyway. AIME is practically useless to 99% of people.Whittemore agrees.At this point, I am fully on the train that these benchmarks are totally soaked, he says. Theres almost no relevant signal, in that all of the models now are at the very high end of these things, and that they just tell you almost nothing.He has this advice for people who are curious about comparable functionality:If you're willing to take the time and the resources to do it, then just try every type of query, and every type of prompt, and every type of challenge, against all of the state of the art (systems) and see which one does best. Or, alternatively, just pick one, assume that it's going to be close to as good as the state of the art, and will be as good as the state of the art in a couple of weeks when they ship the latest update.Anthropics Hybrid ModelLater in the podcast, Whittemore goes over the new Claude 3.7 Sonnet, which he calls a hybrid model based on reasoning and expansive non-reasoning capabilities. Calling the innovation a nudge forward rather than a leap forward, he does concede that SWE-bench improvements and agentic tool use are moved forward with this model.User Reviews of New ModelsFor more, lets turn to a recent post by one of my favorite voices in IT, Ethan Mollick, on his blog, One Useful Thing, and a point that also gets mentioned by Whittemore during the podcast.Mollick has been experimenting with Claude 3.7 Sonnet and Grok 3, and has this to say, in general, about his observations:This new generation of AIs is smarter and the jump in capabilities is striking, particularly in how these models handle complex tasks, math and code, he writes. These models often give me the same feeling I had when using ChatGPT-4 for the first time, where I am equally impressed and a little unnerved by what it can do. Take Claude's native coding ability, I can now get working programs through natural conversation or documents, no programming skill needed.Showing off demos of impressive interactive experiences built with the models, like a time travel simulation thats intuitive, visual, and multi-model, Mollick then talks about two scaling laws that apply:One is that larger models are more capable. Or, as many have observed, we can throw compute at systems and make them work better. The second has to do with test time inference which can also be called inference time compute.OpenAI discovered that if you let a model spend more computing power working through a problem, it gets better results, Mollick writes. (Its) kind of like giving a smart person a few extra minutes to solve a puzzle.Together, these two trends are supercharging AI abilities, and also adding others.The Gen3 generation give the opportunity for a fundamental rethinking of what's possible, he adds. As models get better, and as they apply more tricks like reasoning and internet access, they hallucinate less (though they still make mistakes) and they are capable of higher order thinking.So - less hallucination, better reasoning, more accuracy, more performance, and more propensity for outperforming human PhDs. As Mollick writes: Managers and leaders will need to update their beliefs for what AI can do, and how well it can do it, given these new AI models. Rather than assuming they can only do low-level work, we will need to consider the ways in which AI can serve as a genuine intellectual partner. These models can now tackle complex analytical tasks, creative work, and even research-level problems with surprising sophistication.Theres also an interesting part of the post where Mollick mentions an idea he generated with the new model, a video game based on Herman Melvilles Bartleby, the Scrivener. These are the kinds of projects that will turn heads as we can a view of what AI can now do.Do-It-Yourself AnalysisWhat I hear from all of the above thoughts on AI is that end users should be doing their own research, and figuring out what works best for them.This makes sense, because we have a certain amount of black box issue with LLMs. We dont know exactly how theyre coming to their conclusions. We cant read the actions of digital neurons, obviously. Also, theres quite a bit of subjectivity involved. You can measure model outputs on test sets like GPQA or AIME, but what about for the common things that end users will want to do a teacher planning a lesson plan, an engineer who wants a git push, or a creative professional looking for something for a presentation?Here, a lot of our ratings will be based on real life examples of AI assistance, and not a whole lot of technical benchmarking.