Google Deepmind researchers have built an AI weather forecasting tool that makes faster and more accurate predictions than the best system available today.Dubbed GenCast, the new model outperformed the ENS forecast, widely regarded as the world leader, 97% of the time for predictions up to 15 days in advance. It was tested on over 1,320 weather scenarios, including tropical cyclones and heatwaves.Outperforming ENS marks something of an inflection point in the advance of AI for weather prediction, Ilan Price, a research scientist at Google DeepMind, told the Guardian. At least in the short term, these models are going to accompany and be alongside existing, traditional approaches.GenCast is a diffusion machine learning model, similar to those used in generative AI for tasks like image or text creation. However, its uniquely adapted for weather prediction, trained on four decades of data from the European Centre for Medium-Range Weather Forecasts (ECMWF) the agency behind ENS.The of EU techThe latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!During the experiments, researchers asked GenCast to generate a forecast for 2019. They then compared the results to the actual weather during that year as well as ENS predictions. GenCast creates an ensemble of 50+ different predictions, each showing a possible future scenario. This data helps authorities prepare for extreme weather events like hurricanes or wind farm operators better predict power output days in advance.The fancy name for this technique is probabilistic ensemble forecasting. Its already the gold standard in traditional forecast systems. However, GenCast is taking things up a notch. The system can spit out predictions in far less time: 8 minutes, compared to hours for traditional models.Thats because models like ENS run on massive supercomputers that have to crunch through millions of equations to make a prediction. By contrast, GenCast runs on a single Google Cloud TPU, a chip designed for machine learning. Thats because the AI has been trained, its learnt the data it doesnt have to go through it every single time it needs to make a forecast.GenCast improves upon Deepminds GraphCast model unveiled last year. Other tech firms are also developing their own AI weather forecasters. Nvidia released FourCastNet in 2022, while Huawei launched its Pangu-Weather model in 2023.So will AI replace traditional forecasting soon? Probably not. Models like GenCast still rely on data from traditional weather systems and models to train and calibrate their predictions. However, AI can certainly enhance current methods.The greatest value comes from a hybrid approach, combining human assessment, traditional physics-based models and AI-based weather forecasting, Steven Ramsdale, chief forecaster at the UKs Met Office, told the Financial Times. Story by Sin Geschwindt Sin is a climate and energy reporter at TNW. From nuclear fusion to escooters, he covers the length and breadth of Europe's clean tech ecos (show all) Sin is a climate and energy reporter at TNW. From nuclear fusion to escooters, he covers the length and breadth of Europe's clean tech ecosystem. He's happiest sourcing a scoop, investigating the impact of emerging technologies, and even putting them to the test. Sin has five years journalism experience and holds a dual degree in media and environmental science from the University of Cape Town, South Africa. Get the TNW newsletterGet the most important tech news in your inbox each week.Also tagged with

As holiday shopping continues, Apple is running a fun promotion at Apple Stores this weekend. At select stores in San Francisco, Miami, Los Angeles, and New York City, live artists will be available to personal gift bags for Apple shoppersThe live artsits will be available at these four Apple Stores from 11 a.m. to 3 p.m. local time on Saturday and Sunday: Apple Union Square in San Francisco, CAApple Aventura in Miami, FLApple The Grove in Los Angeles, CAApple Fifth Avenue in NY, NYAt these stores, Apple shoppers can choose to have their gift bags customized by a live artist. You can check out the bios of each artist below. If you head to a participating Apple Store this weekend, be sure to share a picture of your customized gift bag with us in the comments below or on social media!Apple Union Square San FranciscoErik Marinovoich @erikmarinovoichErik Marinovich is a San Francisco based lettering artist and designer, and is a co-founder of Friends of Type. Since 2009, he has drawn letters, logos and type for various consumer brands. In 2012, he co-founded Title Case, a creative work space that conducts workshops and lectures. Between client work, teaching and side-projects, youll find him on the road promoting Keep Fresh Stay Rad and Lets Go Letter Hunting, two new releases from Friends of Type published by Princeton Architectural Press.Apple Aventura MiamiJackson Alves @letterjackJackson Alves is a brazilian calligraphy/lettering artist and educator based in Orlando, Florida. Graduated in graphic design in 2003, in the last twelve years Jackson has been working exclusively as a letterform expert, collaborating with clients from different countries. Alves was awarded the 10th and 12th Graphic Design Biennial in Brazil and also the Certificate of Typographic Excellence by Type Directors Club in NYC. Some of his works are part of the permanent collection of The Contemporary Museum of Calligraphy in Moscow.Apple The Grove Los AngelesKatrina Centeno-Nguyen @calligkatrinaCalligraphy Katrinais a calligraphy firmestablished in 2007 by its highly artistic andtalented founder, Katrina Centeno-Nguyen. Atthe age of 19, she was asked by a friend toaddress her wedding envelopes. With ahumble beginning of $10, Katrina bought apen and a pack of envelopes for samples tostart her first project. After a couple more jobs,she landed a very encouraging write-up in thefamous wedding blog, Project Wedding.Immediately, her natural speed, innate writingskills, and meticulous, yet upbeat personalitywas noticed by Los Angeles luxury weddingindustry, museums, and the worlds topfashion houses including Vogue, Dior,Ferragamo, and Chanelto name afew.Apple Fifth Avenue New YorkLynne Yun @lynneyunLynne Yun is a NYC-based type designer,educator and technologist. Lynne is a founderof Space Type, a studio practice operating atthe intersection of type and technology. Shealso recently co-founded Type Electives, anonline design school shaping the future oftype.Add 9to5Mac to your Google News feed. FTC: We use income earning auto affiliate links. More.Youre 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. Dont know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

I recently wrote about Karo, a new iPhone app from developer Mustafa Yusuf that helps people create to-do lists and assign tasks to others. Starting today, Mac users can also try out Karo with a native version for macOS.Karo task manager now has a Mac appYusuf is also behind Tasks, another to-do app thatIve already written about here on9to5Mac. However, while Tasks is focused on more complex lists and projects, Karo is simpler and aimed at delegating tasks to other people, such as friends and family. And the apps interface is designed to look like youre simply using a chat app.And this is probably the most exciting part of the app, because I often need to remind someone by message about tasks to be done and these messages end up getting lost in long conversations.With Karo, you simply select a contact and initiate a chat. Everything you type creates a new task to be completed. If the other person also has the Karo app, theyll see the task right in the app. But Karo isnt limited to users who have the app installed. For those without it, tasks can be sent via iMessage, WhatsApp, or any other messaging platform you choose.The recipient gets a link to view and mark the task as complete all without needing to install Karo. This makes it especially useful when working with people who dont use iPhones, for example. Once the task is marked complete, youll receive a notification, and the task will update automatically in the app.Karo was first designed to work with the iPhone, but the latest update to the app comes with a Mac version too. Users will find all the same features of the iOS version in the macOS app, including options for setting deadlines and reminders. You can also attach files to each task, use AI to help you manage larger tasks or take a quick look at your to-do list from the Home Screen widgets.Karo is available to try for free on the iOS and Mac App Store. However, after the trial period, it requires either a subscription or a lifetime license. Its worth noting that the subscription is valid for both iOS and macOS versions.Add 9to5Mac to your Google News feed. FTC: We use income earning auto affiliate links. More.Youre 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. Dont know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

Image by Getty / FuturismDevelopmentsThese are scary times to be a multi-millionaire insurance bigwig. The fatal shooting of Brian Thompson, the former CEO of UnitedHealthcare, has c-suites across the country quaking in their boots.As 404 Media reports, with the killer still at larger, major health insurance companies are now taking down their leadership pages and scrubbing any information about their top brass.Almost immediately after Thompson's assassination, a link to UnitedHealthcare's "about us" page was changed so that it no longer directs to a list of the company's executives, 404 found.Similarly, Anthem Blue Cross Blue Shield replaced links to its leadership page with a generic "about us" one with no specific information on its executives. This may be because of the ire it received for its plan, announced around the time of the shooting, to cut off anesthesia coverage for medical procedures that went longer than expected, prompting outrage and viral gallows humor that the assassin's job wasn't over yet. (Blue Cross Blue Shield now says it's axingthe proposed anesthesia plan in certain locations.)404 notes several other health organizations that took similar action, including the nonprofit Caresource, which provides Medicaid and Medicare services; Elevance Health, a for-profit insurance provider part of the Blue Cross Blue Shield group; and Medica, a nonprofit health plan that mainly serves customers in the Midwest. Medica has also temporarily shutdown its headquarters in Minnesota.Based on a company spokesperson's remark, it's no coincidence that major insurers are scrubbing public information."The safety of Medica employees is our top priority and we have increased security both for all of our employees," the spokesperson told FOX 9. "Like other insurers, Medica also has removed from our website the bios of its executives, for the same precautionary reasons."Thompson was fatally shot by an unknown gunman on Wednesday. The assassin's motivations remain unclear, but the words "deny," "defend," and "depose" were reportedly written on the bullet casings recovered from the scene. This is likely a reference to the title of a 2010 book exposing the ruthless methods insurers use to deny healthcare.Echoing that sentiment, the public reaction to Thompson's death has been the opposite of sympathetic,with a wave of outrage at the terrible experiences commentators and their loved ones have had with the insurance industry taking over the discourse. Even doctors have been excoriating insurers in response to the news.With all that in mind, maybe it's not just the gunman at large that these executives are starting to fear.Share This Article

Fast radio bursts (FRBs), hugely powerful discharges of energy emanating from the furthest reaches of outer space, have confounded scientists for well over a decade.They're so powerful that they can release as much energy in a millisecond as the Sun puts out in three entire days, scientists have found.Over the years, researchers have proposed a number of explanations for the unusual phenomenon. One theory that has been put forward is that FRBs are the result of neutron stars, the extremely dense remains of dead stars, colliding with other neutron stars or even black holes.Now, as detailed in a paper that has been accepted for publication in the Astrophysical Journal, a team led by University of Toronto scientist Dang Pham has put a new spin on the concept, proposing the bursts could be caused by asteroids slamming into neutron stars a lesser collision that could still release unfathomable displays of energy.It's an interesting new idea about a phenomenon that has generated dozens of different potential explanations but plenty of work remains until we can get any definitive answers.The team's work is circumstantially intriguing, though: theyestimated the number of interstellar asteroid collisions with neutron stars, and concluded that it appeared to correlate with the estimated number of FRBs observed in the universe."It's been known for many years that asteroids and comets impacting neutron stars can cause FRB-like signals, but until now, it was unclear if this happened often enough across the universe to explain the rate at which we observe FRBs occurring," Pham told Space.com. "We have shown that interstellar objects (ISOs), an understudied class of asteroids and comets thought to be present between stars in galaxies throughout the universe, could be numerous enough that their impacts with neutron stars could explain FRBs!"Previous estimations that take into consideration the two interstellar objects that have been observed so far, Oumuamua in 2017 andComet Borisov in 2019, suggest there could be anywhere from a trillion trillion to 10 trillion trillion that's 26 zeroes similar objects screaming through the Milky Way alone.Scientists know of around 1 billion neutron stars in our galaxy. Taking these figures into account, Pham and his collaborators calculated that one collision could occur every 10 million years per neutron star, which is consistent with the observed rate of FRBs across galaxies.Once the comet or asteroid slams into a neutron star's magnetic field, it could immediately vaporize and speed up "close to the speed of light," as Pham told New Scientist. "This ball of plasma is fired along the magnetic field and creates a beam of radiation that might produce an FRB.""The energy released depends on the size of the asteroid and the strength of the magnetic field on the neutron star, both of which can vary by a lot, by several orders of magnitude," coauthor and Oxford University astrophysicist Matthew Hopkins told Space.com.For an asteroid just over a half mile across, the energy released would be equivalent to "one hundred million times all the energy used by all of humanity over a year," according to Hopkins.The team suggests that if their hypothesis is true, the rate of FRBs could increase as the universe ages.But more work has to be done before we can fully understand these mysterious bursts of energy, from observing more FRBs, calculating how energetic they are, and better understanding how many interstellar objects there are in any given galaxy.Besides, the new hypothesis is far from perfect. For one, there are certain types of FRBs that occur multiple times at the same source and at regular intervals almost like clockwork which don't perfectly align with seemingly random collisions between neutron stars and asteroids, as Universe Today points out.However, it's an intriguing explanation that deserves closer scrutiny as more data continues to pour in.Share This Article

Dec 06, 2024Ravie LakshmananSpyware / Mobile SecurityA Russian programmer accused of donating money to Ukraine had his Android device secretly implanted with spyware by the Federal Security Service (FSB) after he was detained earlier this year.The findings come as part of a collaborative investigation by First Department and the University of Toronto's Citizen Lab."The spyware placed on his device allows the operator to track a target device's location, record phone calls, keystrokes, and read messages from encrypted messaging apps, among other capabilities," according to the report.In May 2024, Kirill Parubets was released from custody after a 15-day period in administrative detention by Russian authorities, during which time his phone, an Oukitel WP7 phone running Android 10, was confiscated from him.During this period, not only was he beaten to compel him into revealing his device password, he was also subjected to an "intense effort" to recruit him as an informant for the FSB, or else risk facing life imprisonment.After agreeing to work for the agency, if only to buy some time and get away, the FSB returned his device at its Lubyanka headquarters. It's at this stage that Parubets began noticing that the phone exhibited unusual behavior, including a notification that said "Arm cortex vx3 synchronization."A further examination of the Android device has since revealed that it was indeed tampered with a trojanized version of the genuine Cube Call Recorder application. It's worth noting that the legitimate app has the package name "com.catalinagroup.callrecorder," whereas the rogue counterpart's package name is "com.cortex.arm.vx3."The counterfeit app is designed to request intrusive permissions that allow it to gather a wide range of data, including SMS messages, calendars, install additional packages, and answer phone calls. It can also access fine location, record phone calls, and read contact lists, all functions that are part of the legitimate app."Most of the malicious functionality of the application is hidden in an encrypted second stage of the spyware," the Citizen Lab said. "Once the spyware is loaded onto the phone and executed, the second stage is decrypted and loaded into memory."The second stage incorporates features to log keystrokes, extract files and stored passwords, read chats from other messaging apps, inject JavaScript, execute shell commands, obtain the device unlock password, and even add a new device administrator.The spyware also exhibits some level of overlap with another Android spyware called Monokle that was documented by Lookout in 2019, raising the possibility that it's either an updated version or that it's been built by reusing Monokle's codebase. Specifically, some of the command-and-control (C2) instructions between the two strains have been found to be identical.The Citizen Lab said it also spotted references to iOS in the source code, suggesting that there could be an iOS version of the spyware."This case illustrates that the loss of physical custody of a device to a hostile security service like the FSB can be a severe risk for compromise that will extend beyond the period where the security services have custody of the device," it said.The disclosure comes as iVerify said it discovered seven new Pegasus spyware infections on iOS and Android devices belonging to journalists, government officials, and corporate executives. The mobile security firm is tracking the spyware developer, NSO Group, as Rainbow Ronin."One exploit from late 2023 on iOS 16.6, another potential Pegasus infection in November 2022 on iOS 15, and five older infections dating back to 2021 and 2022 across iOS 14 and 15," security researcher Matthias Frielingsdorf said. "Each of these represented a device that could have been silently monitored, its data compromised without the owner's knowledge."Found this article interesting? Follow us on Twitter and LinkedIn to read more exclusive content we post.SHARE

Dec 06, 2024Ravie LakshmananArtificial Intelligence / VulnerabilityCybersecurity researchers have disclosed multiple security flaws impacting open-source machine learning (ML) tools and frameworks such as MLflow, H2O, PyTorch, and MLeap that could pave the way for code execution.The vulnerabilities, discovered by JFrog, are part of a broader collection of 22 security shortcomings the supply chain security company first disclosed last month.Unlike the first set that involved flaws on the server-side, the newly detailed ones allow exploitation of ML clients and reside in libraries that handle safe model formats like Safetensors."Hijacking an ML client in an organization can allow the attackers to perform extensive lateral movement within the organization," the company said. "An ML client is very likely to have access to important ML services such as ML Model Registries or MLOps Pipelines."This, in turn, could expose sensitive information such as model registry credentials, effectively permitting a malicious actor to backdoor stored ML models or achieve code execution.The list of vulnerabilities is below -CVE-2024-27132 (CVSS score: 7.2) - An insufficient sanitization issue in MLflow that leads to a cross-site scripting (XSS) attack when running an untrusted recipe in a Jupyter Notebook, ultimately resulting in client-side remote code execution (RCE)CVE-2024-6960 (CVSS score: 7.5) - An unsafe deserialization issue in H20 when importing an untrusted ML model, potentially resulting in RCE A path traversal issue in PyTorch's TorchScript feature that could result in denial-of-service (DoS) or code execution due to arbitrary file overwrite, which could then be used to overwrite critical system files or a legitimate pickle file (No CVE identifier)CVE-2023-5245 (CVSS score: 7.5) - A path traversal issue in MLeap when loading a saved model in zipped format can lead to a Zip Slip vulnerability, resulting in arbitrary file overwrite and potential code executionJFrog noted that ML models shouldn't be blindly loaded even in cases where they are loaded from a safe type, such as Safetensors, as they have the capability to achieve arbitrary code execution."AI and Machine Learning (ML) tools hold immense potential for innovation, but can also open the door for attackers to cause widespread damage to any organization," Shachar Menashe, JFrog's VP of Security Research, said in a statement."To safeguard against these threats, it's important to know which models you're using and never load untrusted ML models even from a 'safe' ML repository. Doing so can lead to remote code execution in some scenarios, causing extensive harm to your organization."Found this article interesting? Follow us on Twitter and LinkedIn to read more exclusive content we post.SHARE

Forrester Principal Analyst Enza Iannopollo explains what the proposed regulations on artificial intelligence actually aim to do, what it means to enterprise CIOs' AI goals, and how to prepare today for the risks and compliance goals of tomorrow.

Enterprise Connect 2025

Showbiz is a tough industry, and making television shows is particularly hard. Theyre much longer than movies, and, if successful, have the potential to become years- and even decades-long commitments for everyone involved, from stars to writers to directors. Its great to be gainfully employed, especially when you have a rabid fanbase clamoring for more as soon as a season of a show is finished, and if your show is popular enough you know youll always have a group of people ready to wait for new installments, no matter how long it takes.For some shows, those gaps of waiting between seasons can be really, really long. The prestige TV era ushered in more intense and more expensive productions, making TV shows that had the same caliber of something youd see in a movie theater, and the cost for all of that was time. But even before the prestige era, shows could vanish for years before making their big comeback a decade or more later, either because of scheduling conflicts or premature cancelations or any number of other reasons.Because these shows are worth the wait, weve counted down the ten series with the longest gaps between seasons, from reality TV to sci-fi epics to dark comedies to underground crime dramas. One thing they all prove is that those who really love something are willing to twiddle their thumbs in the meantime, ready to watch whatever comes out, whenever it comes out. We hated to see these shows go, and were overjoyed when they made their grand return.TV Shows With the Longest Gaps Between SeasonsThese shows took their sweet time going from one season to the next, but dont hold that against them.Gallery Credit: Emma StefanskyREAD MORE: TV Shows People Love That Are Actually BadGet our free mobile appThe Worst Performances By 20 Great ActorsThese actors are all giants in their field. Most have won Oscars for acting! But even with actors this good, they cant all be winners.Categories: Galleries, Lists, Original Features, Special Features, TV News