MacworldThinking about forging a multi-million or even billion-dollar deal with Apple? Hey, weve all been there. Youre sitting around on a Tuesday afternoon, watching old episodes of Columbo in your underwear while eating string cheese thats past its best-by date and you think, I should do a big-time deal with Apple!Well, hold on, Fabio (can the Macalope call you Fabio?). Before you ink that deal, you might want to take a look at how some of the other partners Apples taken on over the years have faired. Because its not always great for the other guy.HPIn 2004, HP started selling special branded iPods. No, really, this happened. You can look it up. HP agreed not to make a rival digital music player through August of 2006 and Apple agreed to not laugh so hard that milk came of its nose. Apple was so into this deal that these iPods had to be repaired by HP, even though they were just standard iPods with an HP logo on them. HPs CEO was soon forced out and the deal was terminated before anyone could figure out what the heck it was all about.Grade: Incomplete, no creditThis photo was not created using AI. Its real.Wikipedia/keeganMotorolaIn 2005, Apple partnered with Motorola on the Rokr, a device that literally everyone agreed was garbage. Limited to 100 songs that could only be transferred to the phone by low-speed USB, the Rokr was eclipsed by the iPod Nano, which was unveiled on the same day.The following year, Motorola unveiled an update that worked with RealPlayer (remember RealPlayer?) instead of iTunes. You can imagine what a hit that was. Apple, meanwhile, was working on the smartphone that would essentially drive Motorola and dozens of other cell phone manufacturers out of business. Only Goldman Sachs could make this deal look slightly better in comparison by losing a metric butt-ton of cash.Grade: DCingular/AT&TApple eased its way into the cellular business without a lot of pain and Cingular got exclusive rights to sell the iPhone in the U.S. for four years. Cingular clearly gave up a lot of control, but it also got something extraordinarily valuable in return. Somewhere Stan Sigman is still monotonously reading from cue cards about this deal, even though he died four years ago.Grade: BIntelUsually, if Apple is buying something from you, its a relatively good relationship (as long as you dont mess anything up). Foxconn has done alright. So has Samsung. Intel did fine for years, but it really missed the memo on processors for mobile devices.Most of this is the fault of Intel itself, but Apple was also quietly improving its iPhone processors so they could be used in Macs to run fast with very low power consumption. Intel was sailing along as if nothing had changed since 1997.Grade: C-GoogleGoogle has largely done fine in its relationship with Apple, you know, other than getting into an argument about Maps and having to deal with lawsuits about Android. Its lucrative search deal has drawn the attention of the U.S. Department of Justice, but its likely that some well-time donations to the incoming administrations inauguration will snuff that out. Definitely a passing grade, if not terribly high.Grade: B-Goldman SachsGoldman Sachs partnered with Apple on the Apple Card and, boy, does it wish it had not done that. How do you lose $859 million in a year? Who does that? Probably a company thats used to having the taxpayers bail it out, thats who.Goldmans mistake here seems to have been rushing so fast to get into the consumer business that it didnt A) ask itself it was capable of getting into the consumer business and B) read the fine print on its deal with Apple. Something something ground war in Asia, something something going in against a Sicilian when death is on the line.Grade: FNikeApple has been selling Nike-themed Watches and bands for years now and this relationship largely works because Apple and Nike are in completely different lines of business (same with Herms). You might think, hey, so are Apple and Goldman Sachs! But, no, not according to the U.S. government.Grade: A-The Macalope isnt saying you cant win by partnering with Apple. Clearly you can do fine and there are, of course, many non-monetary benefits to partnering with an 800-lb gorilla. But, you should expect that the gorilla is going to get more of the bananas than you are. If you go into the relationship thinking, hey, lotta bananas to be had by partnering with this gorilla because people keep throwing it bananas, think again. There are a lot of bananas that will be thrown, yes, but youll mostly get left with the peels and maybe those little stringy bits no one likes.By the way, there are no flaws in this gorilla analogy. None.

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.Peering into the body to find and diagnose cancer is all about spotting patterns. Radiologists use x-rays and magnetic resonance imaging to illuminate tumors, and pathologists examine tissue from kidneys, livers, and other areas under microscopes and look for patterns that show how severe a cancer is, whether particular treatments could work, and where the malignancy may spread.In theory,artificial intelligence should be great at helping out. Our job is pattern recognition, says Andrew Norgan, a pathologist and medical director of the Mayo Clinics digital pathology platform. We look at the slide and we gather pieces of information that have been proven to be important.Visual analysis is something that AI has gotten quite good at since the first image recognition models began taking off nearly 15 years ago. Even though no model will be perfect, you can imagine a powerful algorithm someday catching something that a human pathologist missed, or at least speeding up the process of getting a diagnosis. Were starting to see lots of new efforts to build such a modelat least seven attempts in the last year alonebut they all remain experimental. What will it take to make them good enough to be used in the real world?Details about the latest effort to build such a model, led by the AI health company Aignostics with the Mayo Clinic, were published on arXiv earlier this month. The paper has not been peer-reviewed, but it reveals much about the challenges of bringing such a tool to real clinical settings.The model, called Atlas, was trained on 1.2 million tissue samples from 490,000 cases. Its accuracy was tested against six other leading AI pathology models. These models compete on shared tests like classifying breast cancer images or grading tumors, where the models predictions are compared with the correct answers given by human pathologists. Atlas beat rival models on six out of nine tests. It earned its highest score for categorizing cancerous colorectal tissue, reaching the same conclusion as human pathologists 97.1% of the time. For another task, thoughclassifying tumors from prostate cancer biopsiesAtlas beat the other models high scores with a score of just 70.5%. Its average across nine benchmarks showed that it got the same answers as human experts 84.6% of the time.Lets think about what this means. The best way to know whats happening to cancerous cells in tissues is to have a sample examined by a pathologist, so thats the performance that AI models are measured against. The best models are approaching humans in particular detection tasks but lagging behind in many others. So how good does a model have to be to be clinically useful?Ninety percent is probably not good enough. You need to be even better, says Carlo Bifulco, chief medical officer at Providence Genomics and co-creator of GigaPath, one of the other AI pathology models examined in the Mayo Clinic study. But, Bifulco says, AI models that dont score perfectly can still be useful in the short term, and could potentially help pathologists speed up their work and make diagnoses more quickly.What obstacles are getting in the way of better performance? Problem number one is training data.Fewer than 10% of pathology practices in the US are digitized, Norgan says. That means tissue samples are placed on slides and analyzed under microscopes, and then stored in massive registries without ever being documented digitally. Though European practices tend to be more digitized, and there are efforts underway to create shared data sets of tissue samples for AI models to train on, theres still not a ton to work with.Without diverse data sets, AI models struggle to identify the wide range of abnormalities that human pathologists have learned to interpret. That includes for rare diseases, says Maximilian Alber, cofounder and CTO of Aignostics. Scouring the publicly available databases for tissue samples of particularly rare diseases, youll find 20 samples over 10 years, he says.Around 2022, the Mayo Clinic foresaw that this lack of training data would be a problem. It decided to digitize all of its own pathology practices moving forward, along with 12 million slides from its archives dating back decades (patients had consented to their being used for research). It hired a company to build a robot that began taking high-resolution photos of the tissues, working through up to a million samples per month. From these efforts, the team was able to collect the 1.2 million high-quality samples used to train the Mayo model.This brings us to problem number two for using AI to spot cancer. Tissue samples from biopsies are tinyoften just a couple of millimeters in diameterbut are magnified to such a degree that digital images of them contain more than 14 billion pixels. That makes them about 287,000 times larger than images used to train the best AI image recognition models to date.That obviously means lots of storage costs and so forth, says Hoifung Poon, an AI researcher at Microsoft who worked with Bifulco to create GigaPath, which was featured in Nature last year. But it also forces important decisions about which bits of the image you use to train the AI model, and which cells you might miss in the process. To make Atlas, the Mayo Clinic used whats referred to as a tile method, essentially creating lots of snapshots from the same sample to feed into the AI model. Figuring out how to select these tiles is both art and science, and its not yet clear which ways of doing it lead to the best results.Thirdly, theres the question of which benchmarks are most important for a cancer-spotting AI model to perform well on. The Atlas researchers tested their model in the challenging domain of molecular-related benchmarks, which involves trying to find clues from sample tissue images to guess whats happening on a molecular level. Heres an example: Your bodys mismatch repair genes are of particular concern for cancer, because they catch errors made when your DNA gets replicated. If these errors arent caught, they can drive the development and progression of cancer.Some pathologists might tell you they kind of get a feeling when they think somethings mismatch-repair deficient based on how it looks, Norgan says. But pathologists dont act on that gut feeling alone. They can do molecular testing for a more definitive answer. What if instead, Norgan says, we can use AI to predict whats happening on the molecular level? Its an experiment: Could the AI model spot underlying molecular changes that humans cant see?Generally no, it turns out. Or at least not yet. Atlass average for the molecular testing was 44.9%. Thats the best performance for AI so far, but it shows this type of testing has a long way to go.Bifulco says Atlas represents incremental but real progress. My feeling, unfortunately, is that everybodys stuck at a similar level, he says. We need something different in terms of models to really make dramatic progress, and we need larger data sets.Now read the rest of The AlgorithmDeeper LearningOpenAI has created an AI model for longevity scienceAI has long had its fingerprints on the science of protein folding. But OpenAI now says its created a model that can engineer proteins, turning regular cells into stem cells. That goal has been pursued by companies in longevity science, because stem cells can produce any other tissue in the body and, in theory, could be a starting point for rejuvenating animals, building human organs, or providing supplies of replacement cells.Why it matters: The work was a product of OpenAIs collaboration with the longevity company Retro Labs, in which Sam Altman invested $180 million. It represents OpenAIs first model focused on biological data and its first public claim that its models can deliver scientific results. The AI model reportedly engineered more effective proteins, and more quickly, than the companys scientists could. But outside scientists cant evaluate the claims until the studies have been published. Read more from Antonio Regalado.Bits and BytesWhat we know about the TikTok banThe popular video app went dark in the United States late Saturday and then came back around noon on Sunday, even as a law banning it took effect. (The New York Times)Why Meta might not end up like XX lost lots of advertising dollars as Elon Musk changed the platforms policies. But Facebook and Instagrams massive scale make them hard platforms for advertisers to avoid. (Wall Street Journal)What to expect from Neuralink in 2025More volunteers will get Elon Musks brain implant, but dont expect a product soon. (MIT Technology Review)A former fact-checking outlet for Meta signed a new deal to help train AI modelsMeta paid media outlets like Agence France-Presse for years to do fact checking on its platforms. Since Meta announced it would shutter those programs, Europes leading AI company, Mistral, has signed a deal with AFP to use some of its content in its AI models. (Financial Times)OpenAIs AI reasoning model thinks in Chinese sometimes, and no one really knows whyWhile it comes to its response, the model often switches to Chinese, perhaps a reflection of the fact that many data labelers are based in China. (Tech Crunch)

Gusts Atelier Yumia: The Alchemist of Memories and the Envisioned Land is two months off, but theres still more to reveal. Thankfully, a new broadcast for the title will go live on January 22nd at this years Taipei Game Show at 3 AM PT. Producer Junzo Hosoi will be present with actors Wakana Kuramochi (Yumia) and Kaori Maeda (Isla). Live gameplay with Chinese localization will be presented, so stay tuned for new information and details.Atelier Yumia: The Alchemist of Memories and the Envisioned Land launches on March 21st for Xbox Series X/S, Xbox One, PS4, PS5, PC, and Nintendo Switch. Players control Yumia Liessfeldt as she investigates the Aladissian Empires collapse due to alchemy. Youll still craft items from raw materials, though players can create them on the go, and combat occurs on the field in real time. Check out the latest trailer, featuring the antagonists opposing Yumia and her friends. You can also learn more about how Halo inspired the open-world design.

Arc System Works has announced the release date for Granblue Fantasy Versus: Risings next DLC character, Sandalphon. Hes available on February 26th for $7.99 standalone or as part of Character Pass 2. Check out the latest trailer showcasing his abilities below.Opposing Lucilius, Sandalphon is more than 2000 years old. As the Primarch of Promises, he possesses incredible powers despite mostly fighting with a one-handed sword. These include elemental projectile attacks, from direct flaming shots to earthen spikes from below. His multiple wings also come into play, though its unknown whether this is a powered-up state or for his Super Skybound Art.Stay tuned for more details on Sandalphon and his move set in the coming weeks. Cygames also confirmed release windows for the remaining fighters in Character Pass 2. The next debuts in Spring 2025, followed by the next in Summer 2025, then Fall 2025, and the final DLC fighter in early 2026.Granblue Fantasy Versus: Rising is available for PS4, PS5, and PC. Check out our review for the base game here.

html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd"Epic Games has posted a recording of the session on upcoming features in Unreal Engine, its game engine and real-time renderer, from last years Unreal Fest Seattle conference.It runs through features in development for Unreal Engine 5.6, the next version of the software, along with those in development for Unreal Engine 5.7 and beyond.The video covers changes to gameplay systems, audio, build tools and platform support, but below, weve summarised the key changes affecting CG artists, as opposed to programmers.They span UE5s world-building, rendering, animation and simulation toolsets, and range from concepts like Megaworlds biomes to big initiatives like the Anim Next unified animation pipeline.New features scheduled for Unreal Engine 5.6 and beyondIn the recording, Epic Games Technical Director Arjan Brussee runs through upcoming features in Unreal Engine.The session took place a month before the release of Unreal Engine 5.5, so many of the features shown have already been released, but a number are still in development.Some are scheduled for Unreal Engine 5.6, which is likely to be released in spring or early summer this year, while others arent yet scheduled for specific releases.However, theres a big disclaimer at the start of the video that all of the plans announced are tentative, so when or if individual features ship could change at any time.Nanite: Updates to Nanite FoliageNanite Foliage that is, support for assets like trees inside Nanite, Unreal Engine 5s geometry-streaming system, is due for some really big updates in Unreal Engine 5.6.The slide above namechecks Nanite decals & translucent support: presumably support for Mesh Decals using a Translucent Blend Mode within Nanite.The work is intended to resolve issues with rendering dynamically moving trees, and to make Nanite Foliage work better with Lumen, Unreal Engines dynamic global illumination system.Lumen: 120Hz mode and support for low-end hardwareWork on Lumen itself focuses on improving performance, with Epic working on a 120Hz mode up from 60Hz on current consoles and a prototype capable of running on low-end hardware.The MegaLights system the Nanite of lights introduced in Unreal Engine 5.5 will also see continued R&D work.Rendering: Better rendering of first-person gamesRendering changes include better rendering of the player mesh in first-person games.The work is intended to reduce the need for nasty hacks to render the mesh, and includes better support for self-shadowing and reflections, and a separate FOV for first-person mode.Rendering: More intuitive new material editorMany of the other changes announced to rendering are iterative improvements, including further reducing shader permutations by reducing the need for Static Switch conversions.However, Brussee also namechecked an experimental new Material Editor intended to make it easier for non-experts to make performant materials simply out of the box.World Creation: World Partition Bookmarks and better streaming performanceChanges due to the world creation toolset in Unreal Engine 5.6 include World Partition Bookmarks, intended to help artists navigate large open worlds.The update should also improve performance when streaming data in and out of memory, particularly on complex environments with tens of thousands of objects in a cell.Longer-term changes include more automated tools for packing objects into grid cells, and implementation of the Scene Graph from Unreal Editor for Fortnite (UEFN).PCG: Megaworlds biomes simplify environment creationUnreal Engines Procedural Content Generation Framework (PCG) will get workflow improvements, including better attribute support, dependency tracking and versioning.However, Brussee also announced a new system of Megaworlds biomes: assets with built-in PCG rules that could be released on online marketplaces like Fab to speed up world building.The slide above also lists PCG for landscape grass, described in Epics online roadmap as a GPU compute-based solution for landscape and micro-scattering for high-frequency details.Terrain: next-gen terrain system will bring Nanite-level detail to landscape geometryEpic Games is also working on a next-generation terrain solution, potentially bringing Unreal Engines currently heightfield-only landscape system into full 3D.Brussee noted that people internally sometimes say that our landscape is the worst geometry we have on screen, and that the new system was intended to get terrain to Nanite level detail.Character Rigging: Physics Rigs and a new Skeletal Hair Editor Changes to Control Rig, Unreal Engines character rigging and animation system, include updates to the Deformer Graph, and new nodes for creating Physics Rigs of which, more later.The slide above also mentions a new Skeletal Hair Editor, although it wasnt mentioned in the presentation, and isnt yet listed on the online roadmap.Animation: New branching dialogue systemSequencer, Unreal Engines cinematic editor, is scheduled to get quite a few new features, including support for animation mixing with bone masking.However, perhaps the biggest is a full branching dialogue system, out of the box.According to the online roadmap, the Narrative Branching with Cinematic Dialogues system will help artists create branching cinematics, including procedural content-generation capabilities.Animation: in-engine animation with character physicsChanges to Unreal Engines in-engine animation tools will include a simplified timeline, and new versions of the Tween Tools and Motion Trails systems.The one discussed in detail was animating with physics, with a physics sim run in parallel with skeletal animation, which Brussee described as providing free overlap and secondary motion.Animation: support for multi-actor Motion MatchingOther animation changes include multi-character Motion Matching, making it possible to have character controlled by the Motion Matching motion synthesis system interact.The experimental Gameplay Camera System will also move into beta, while the Mutable plugin for generating character variations will become production-ready.Anim Next: new high-performance unified animation pipelineHowever, perhaps the biggest upcoming change to animation in Unreal Engine is Anim Next: a long-term initiative to integrate all of the animation logic systems.The new unified animation pipeline will be faster and easier to use than Animation Blueprints, which it is ultimately intended to replace, removing the need to nativize Blueprints into C++.Chaos Physics: new machine learning capabilitiesOver in Unreal Engines Chaos Physics toolset, which includes soft body system Chaos Flesh, Epic plans to fully automate musculoskeletal simulation inside the [Unreal] Editor itself.The slide above also namechecks new machine learning features, following the ML Deformer introduced in Unreal Engine 5.2, including machine-learned SDFs and contact generation.Niagara: integrate Niagara effects with the PCGThe planned changes to Niagara, Unreal Engines particle-based VFX system, look to be mainly workflow improvements, with Brussee noting that the toolset had become too expert-level.However, Epic also plans to make Niagara and the PCG interoperable, making it possible to have simulations of different types inform and affect one another.Price, system requirements and release dateUnreal Engine 5.6 due later in 2025. Epic Games hasnt announced an exact date. The current release, Unreal Engine 5.5, is available for 64-bit Windows, macOS and Linux.For non-interactive content, the software is free to users with revenue under $1 million/year. For larger studios, subscriptions cost $1,850/seat/year, including Twinmotion and RealityCapture.For games developed with the engine, Epic takes 5% of the gross royalties after the first $1 million generated.See features currently in development for Unreal Engine on Epic Games public roadmapHave your say on this story by following CG Channel on Facebook, Instagram and X (formerly Twitter). As well as being able to comment on stories, followers of our social media accounts can see videos we dont post on the site itself, including making-ofs for the latest VFX movies, animations, games cinematics and motion graphics projects.

I have an AirTag on my keychain to keep track of my keys, but obviously an AirTag is too big and bulky to fit inside a wallet. Thats what the SwitchBot Wallet Finder is for. Disguised inside a thin, credit-card form factor, the SwitchBot Wallet Finder connects to the Find My app on your iPhone, so you can follow its location, and it even houses a speaker so you can make it beep to help you find your wallet when it inevitably gets lost somewhere in your home.Read on for my review SetupSetup is super simple. Although it ostensibly supports its own app and Android, I have used it solely through the Apple Find My app built into the phone. I didnt even download the SwitchBot app at all. To set up, you open Find My, tap Add Item, and press and hold the button on the Wallet Finder card for a couple of seconds. You then choose a name and emoji icon to identify it later. By default, it suggested Benjamins Keys, but I renamed it to Benjamins Wallet, selected one of the wallet-adjacent available emojis, and pressed Continue and I was done. Then, it just slips into one of the empty card pockets in my wallet.How it tracks locationAs a reminder, Find My accessories do not have GPS. Instead, they broadcast a low-energy Bluetooth signal that nearby Apple devices like iPhones, iPads and Macs pick up on. The location is then transmitted securely and privately to the Find My network, so you can hopefully locate and retrieve your lost item.Even if you are miles away, as long as someone with an Apple device is nearby, the wallet will be locatable on the map in the Find My app.Theres also no ongoing subscription or other costs to worry about (unlike some competitors).As demonstrated by the success of AirTags, the Find My network works really well and the SwitchBot Wallet Finder is just as effective. If your item is stationary, and nearby one of the billion iOS devices in the world, it wont take too long for its location to appear in Find My. If the item is moving, location updates are slightly delayed behind real-time, however.You can also use the Find My app to mark your item as explicitly lost. Then, if someone else finds your wallet and wants to help give it back to its rightful owner, they can use the Identify Found Item feature in Find My which will present your contact information to help arrange returning it to you. The upcoming iOS 18.2 update even adds the ability to make a shareable link to send to others to help assist in tracking down your lost items.Finding nearby using the speakerUnlike an AirTag, the Wallet Finder does not incorporate an ultra-wide band radio. That means it lacks the Precision Finding feature that the AirTag offers when in close proximity.However, the integrated speaker in the Wallet Finder more than suffices for the job of finding your lost wallet. When in Bluetooth range, the card emits a reasonably loud tone. It is slightly less loud than the sound an AirTag makes, and naturally gets a bit muffed when ensconced inside a closed wallet, but it is still audible, more than enough for you to hear it from a good distance away.For the (embarrassingly common) case of when Ive misplaced my wallet somewhere inside my house, I can now just open the Find My app, press Play Sound, and instantly hear which room it is in.Find My also works with the Siri voice assistant, as an added convenience. For example, I can shout wheres my wallet? at my HomePod, and it will start pinging it for me.Safety and trackingIve been using Pebblebee Card for the last couple of years and I love it! They released a new version recently that has even longer battery life. Rechargeable also! View all commentsJust like AirTags, the Wallet Finder is marketed as a device to help recover lost items. It is not designed to address problems of theft or nefariously track other people. If the card is separated from its owner for a period of time, the speaker will emit a noise to make people aware of its presence. It will also trigger Unknown Item Moving With You alerts on any nearby iPhones in the vicinity.Its also worth pointing out that someone else with the card in their possession can deactivate the Find My connection, by pressing the button in a particular shutdown sequence.ConclusionIf used for its intended purpose, the SwitchBot Wallet Finder is a very effective and great value for money. While obviously designed to be kept inside a wallet, it even has a built-in hole cutout if you did want to slip it on some kind of lanyard or keychain. I highly recommend it.Perhaps the only drawback to be aware of is that the Wallet Finder card has no means of replacing the battery. It is a single use consumable, rated for about 3 years of usage. But the contents of your wallet are valuable enough that it probably makes financial sense to simply buy another when it eventually dies. You just have to be okay with yourself regarding the environmental waste.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

Jan 21, 2025The Hacker NewsImagine receiving a penetration test report that leaves you with more questions than answers. Questions like, "Were all functionalities of the web app tested?" or " Were there any security issues that could have been identified during testing?" often go unresolved, raising concerns about the thoroughness of the security testing. This frustration is common among many security teams. Pentest reports, while crucial, frequently lack the depth and detail necessary to truly assess the success of the project.Even with years of experience working with cybersecurity teams and managing ethical hacking projects, we frequently encountered these same issues. Whether collaborating with external pentest providers or managing our own projects as founders of Hackrate, we often faced difficulties in ensuring that the testing was as comprehensive as it needed to be.This realization inspired us to create HackGATE, a managed gateway solution built to bring transparency and control to pentesting projects, ensuring no questions are left unanswered about the quality and thoroughness of the penetration test projects. We aimed to not only address our own challenges but also to provide the cybersecurity industry with a powerful tool to enhance visibility in their ethical hacking projects.Common Challenges in Penetration Testing1. Lack of visibility and controlA recent survey on pentest projects revealed that 60% of security professionals struggle to measure the success of their pentests. Additionally, nearly two-thirds (65%) of respondents rely solely on information provided by the pentest vendor. This highlights a significant gap in the cybersecurity landscape: the lack of a solution offering visibility into pentesting activities. Without such a solution, security teams struggle with limited insight into crucial aspects of the testing process, including the overall scope and duration of the tests, the specific techniques and attack vectors employed, and the detailed steps taken by ethical hackers.2. Dependence on the final pentest reportMost companies that outsource pentests depend on a final report and their trust in the pentest vendor to assess success. Without concrete evidence of the various aspects of the testing, security teams are left with concerns and security blind spots, encountering obstacles both in understanding their security testing projects and in communicating their outcomes to leadership and stakeholders.3. Coordination in remote pentester teamsManaging a globally distributed team, particularly when working across different time zones, adds to these challenges. This can lead to delays in communication and coordination, resulting in missed deadlines and incomplete tasks. Ensuring that all team members adhere to the same standards across various locations is also challenging. Inconsistent practices can lead to gaps in pentest coverage, leaving critical vulnerabilities undiscovered.How HackGATE Addresses These Challenges1. Enhanced visibility and detailed insightsHackGATE provides real-time visibility into pentest activities. For instance, it details the security testing traffic sent to targets, highlights targeted testing areas, and outlines the methods used by ethical hackers. This transparency ensures you can track the security testing process effectively. 2. Establishing a quality framework for ethical hackingTo ensure the quality of the testing process, it is crucial to establish controls based on analyzed data. Ethical hackers use guidelines and best practices, such as the OWASP guidelines, to provide a structured approach to identifying security risks. While OWASP's framework offers a thorough evaluation of web applications, auditing the security tests is still necessary to verify that pentesters are truly following the guidelines.HackGATE ensures the effectiveness of penetration tests by establishing baselines for minimum testing traffic, which includes both manual and automated testing activities. This ensures thoroughness and consistency in assessments.3. Consolidated and visualized dataPenetration tests generate large volumes of data, which can be difficult to analyze and understand with traditional Security Operation Center solutions. Teams need a centralized dashboard that consolidates key insights, showing the most important metrics, so all stakeholders can easily keep up with progress and monitor ethical hacking activities.HackGATE's unified dashboard addresses this need by consolidating critical insights into a single view. It includes features for project management, analytics, and a detailed overview of pentester activities. This allows all stakeholders to easily access and understand the key metrics without sifting through disparate sources.4. Better coordination across distributed security teamsBy providing a unified interface for all team members, HackGATE ensures that everyone adheres to the same standards, reducing inconsistencies in pentest coverage. The platform also supports comprehensive scope coverage by enabling accurate and detailed reporting, ensuring that all intended assets are tested and documented.HackGATE also enhances accountability by automatically generating detailed reports, providing evidence of testing. This not only helps in holding team members accountable but also simplifies the audit process, ensuring regulatory compliance with a clear and accessible audit trail.HackGATE approachTo ensure successful penetration testing initiatives, security teams need to adopt the 'Trust but Verify' principle in penetration test. This means that instead of relying solely on their pentest provider's report, they need to be able to verify the quality and thoroughness of the testing. But how can they achieve this? The 'Trust but Verify' approach requires accurate data, effective monitoring, and detailed reporting. Most companies still struggle due to the lack of methodology and tools.ConclusionTo ensure your penetration testing projects are comprehensive and compliant, consider integrating innovative monitoring tools like HackGATE into your cybersecurity strategy. For a more in-depth understanding of how it can address your specific needs, schedule a consultation with our technical experts - no sales pitch, just a detailed exploration of how our solution can enhance your pentest approach. Visit the HackGATE website to get started or arrange your personalized technical consultation.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.SHARE

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Peering into the body to find and diagnose cancer is all about spotting patterns. Radiologists use x-rays and magnetic resonance imaging to illuminate tumors, and pathologists examine tissue from kidneys, livers, and other areas under microscopes and look for patterns that show how severe a cancer is, whether particular treatments could work, and where the malignancy may spread. In theory,artificial intelligence should be great at helping out. Our job is pattern recognition, says Andrew Norgan, a pathologist and medical director of the Mayo Clinics digital pathology platform. We look at the slide and we gather pieces of information that have been proven to be important. Visual analysis is something that AI has gotten quite good at since the first image recognition models began taking off nearly 15 years ago. Even though no model will be perfect, you can imagine a powerful algorithm someday catching something that a human pathologist missed, or at least speeding up the process of getting a diagnosis. Were starting to see lots of new efforts to build such a modelat least seven attempts in the last year alonebut they all remain experimental. Details about the latest effort to build such a model, led by the AI health company Aignostics with the Mayo Clinic, were published on arXiv earlier this month. The paper has not been peer-reviewed, but it reveals much about the challenges of bringing such a tool to real clinical settings. The model, called Atlas, was trained on 1.2 million tissue samples from 490,000 cases. Its accuracy was tested against six other leading AI pathology models. These models compete on shared tests like classifying breast cancer images or grading tumors, where the models predictions are compared with the correct answers given by human pathologists. Atlas beat rival models on six out of nine tests. It earned its highest score for categorizing cancerous colorectal tissue, reaching the same conclusion as human pathologists 97.1% of the time. For another task, thoughclassifying tumors from prostate cancer biopsiesAtlas beat the other models high scores with a score of just 70.5%. Its average across nine benchmarks showed that it got the same answers as human experts 84.6% of the time. Lets think about what this means. The best way to know whats happening to cancerous cells in tissues is to have a sample examined by a pathologist, so thats the performance that AI models are measured against. The best models are approaching humans in particular detection tasks but lagging behind in many others. So how good does a model have to be to be clinically useful? Ninety percent is probably not good enough. You need to be even better, says Carlo Bifulco, chief medical officer at Providence Genomics and co-creator of GigaPath, one of the other AI pathology models examined in the Mayo Clinic study. But, Bifulco says, AI models that dont score perfectly can still be useful in the short term, and could potentially help pathologists speed up their work and make diagnoses more quickly. What obstacles are getting in the way of better performance? Problem number one is training data. Fewer than 10% of pathology practices in the US are digitized, Norgan says. That means tissue samples are placed on slides and analyzed under microscopes, and then stored in massive registries without ever being documented digitally. Though European practices tend to be more digitized, and there are efforts underway to create shared data sets of tissue samples for AI models to train on, theres still not a ton to work with. Without diverse data sets, AI models struggle to identify the wide range of abnormalities that human pathologists have learned to interpret. That includes for rare diseases, says Maximilian Alber, cofounder and CTO of Aignostics. Scouring the publicly available databases for tissue samples of particularly rare diseases, youll find 20 samples over 10 years, he says. Around 2022, the Mayo Clinic foresaw that this lack of training data would be a problem. It decided to digitize all of its own pathology practices moving forward, along with 12 million slides from its archives dating back decades (patients had consented to their being used for research). It hired a company to build a robot that began taking high-resolution photos of the tissues, working through up to a million samples per month. From these efforts, the team was able to collect the 1.2 million high-quality samples used to train the Mayo model. This brings us to problem number two for using AI to spot cancer. Tissue samples from biopsies are tinyoften just a couple of millimeters in diameterbut are magnified to such a degree that digital images of them contain more than 14 billion pixels. That makes them about 287,000 times larger than images used to train the best AI image recognition models to date. That obviously means lots of storage costs and so forth, says Hoifung Poon, an AI researcher at Microsoft who worked with Bifulco to create GigaPath, which was featured in Nature Thirdly, theres the question of which benchmarks are most important for a cancer-spotting AI model to perform well on. The Atlas researchers tested their model in the challenging domain of molecular-related benchmarks, which involves trying to find clues from sample tissue images to guess whats happening on a molecular level. Heres an example: Your bodys mismatch repair genes are of particular concern for cancer, because they catch errors made when your DNA gets replicated. If these errors arent caught, they can drive the development and progression of cancer. Some pathologists might tell you they kind of get a feeling when they think somethings mismatch-repair deficient based on how it looks, Norgan says. But pathologists dont act on that gut feeling alone. They can do molecular testing for a more definitive answer. What if instead, Norgan says, we can use AI to predict whats happening on the molecular level? Its an experiment: Could the AI model spot underlying molecular changes that humans cant see? Generally no, it turns out. Or at least not yet. Atlass average for the molecular testing was 44.9%. Thats the best performance for AI so far, but it shows this type of testing has a long way to go. Bifulco says Atlas represents incremental but real progress. My feeling, unfortunately, is that everybody's stuck at a similar level, he says. We need something different in terms of models to really make dramatic progress, and we need larger data sets. Now read the rest of The Algorithm Deeper Learning OpenAI has created an AI model for longevity science AI has long had its fingerprints on the science of protein folding. But OpenAI now says its created a model that can engineer proteins, turning regular cells into stem cells. That goal has been pursued by companies in longevity science, because stem cells can produce any other tissue in the body and, in theory, could be a starting point for rejuvenating animals, building human organs, or providing supplies of replacement cells. Why it matters: The work was a product of OpenAIs collaboration with the longevity company Retro Labs, in which Sam Altman invested $180 million. It represents OpenAIs first model focused on biological data and its first public claim that its models can deliver scientific results. The AI model reportedly engineered more effective proteins, and more quickly, than the companys scientists could. But outside scientists cant evaluate the claims until the studies have been published. Read more from Antonio Regalado. Bits and Bytes What we know about the TikTok ban The popular video app went dark in the United States late Saturday and then came back around noon on Sunday, even as a law banning it took effect. (The New York Times) Why Meta might not end up like X X lost lots of advertising dollars as Elon Musk changed the platform's policies. But Facebook and Instagrams massive scale make them hard platforms for advertisers to avoid. (Wall Street Journal) What to expect from Neuralink in 2025 More volunteers will get Elon Musks brain implant, but dont expect a product soon. (MIT Technology Review) A former fact-checking outlet for Meta signed a new deal to help train AI models Meta paid media outlets like Agence France-Presse for years to do fact checking on its platforms. Since Meta announced it would shutter those programs, Europes leading AI company, Mistral, has signed a deal with AFP to use some of its content in its AI models. (Financial Times) OpenAIs AI reasoning model thinks in Chinese sometimes, and no one really knows why While it comes to its response, the model often switches to Chinese, perhaps a reflection of the fact that many data labelers are based in China. (Tech Crunch)

Our Experts Written by Giselle Castro-Sloboda Our expert, award-winning staff selects the products we cover and rigorously researches and tests our top picks. If you buy through our links, we may get a commission. Reviews ethics statement How we test Table of Contents What to consider Your exercise routine The first thing to consider is what type of exercise you want to be doing at home; is it for strength training, cardio or stress relief? Your answer (or answers) will dictate which pieces you should invest in. Space How much space you have to store things will matter when you're considering what home exercise equipment to buy. A yoga mat is going to take up less space than an elliptical in your living room, for example. What are you actually going to use? You don't want to invest time and living space in something you'll only pick up once or twice a year. Choose equipment you can see yourself using regularly. Having your own gym at home has more pros than cons. For one thing, you can work out whenever you want, and secondly, it's completely free. But what is the best home equipment for your gym if you want the very best workout? To find the answer, I looked at factors like size, versatility, fitness level, and overall cost. I also drew from my personal experience as a trainer, listing tools and machines I used with clients -- all of which I continue to use in my own workouts at home. I've hand-tested each piece of equipment listed here, so I know which equipment is the best. Read on to find out which home gym gear is suitable for you, from the best dumbbells to the best cardio machines. And if you're looking for ways to make your home gym smarter, we've got just the list for that, too. Our Picks Best home exercise equipment for low-impact exercises Yoga mat View details $26 at Amazon View details Best weights for home exercise equipment Dumbbells View details $49 at Walmart View details Best home exercise equipment that can be used while traveling Resistance bands View details $199 at LIT View details Best home exercise equipment to challenge bodyweight Suspension training system View details $60 at Amazon View details Best home exercise equipment to challenge weight bearing exercises Adjustable weight bench View details $150 at Amazon View details Best home exercise equipment that can be used for cardio and strength training Kettlebells View details $42 at Roguefitness View details Best affordable cardio home exercise equipment Jump rope View details $7 at Amazon View details Best home exercise equipment to challenge the core Medicine Ball View details $35 at Walmart View details Table of Contents What is the best home exercise equipment? The ideal piece of home exercise equipment depends on your preferences, goals, and available space for a home gym. Some pieces of exercise equipment are big, like treadmills, so they require plenty of room, whereas others like dumbbells are small enough to store off to the side either on the floor or in a weight rack, if you're fancy. Best home exercise equipment for 2025 Photo Gallery 1/1 $26 at Amazon Pros Easy to store Doesn't take up much space Cons Relies on body weight exercises or addition of other equipment for strength workouts $26 at Amazon A yoga mat is a key piece of equipment to have around. You can use it for low-impact floor exercises, such as Pilates and yoga, or for boot camps and other higher-impact exercises. Shop with the high-quality brandManduka or an inexpensive one such as Gaiam, which offers non-slip options for all types of workouts. To prolong the life of your yoga mat, make sure to clean it regularly by following CNET's yoga mat cleaning guide.

Key Takeaways The average savings rate is still above 4% APY, but only one bank has a 5% APY.Many banks have lowered savings rates slightly this month ahead of the Federal Reserve's meeting.Now's a good time to take advantage of high-yield savings accounts, many of which don't require a minimum balance or deposit to earn interest. If you've set financial goals this year, like building up an emergency fund or saving money to move, the best high-yield savings accounts can help. As with a traditional savings account, you can regularly deposit and withdraw money when you need it, but you can earn more interest on your balance.Annual percentage yields for the best high-yield savings accounts are currently over 4%, which is nearly 10 times higher than the national average of 0.42% APY.For now, savings account rates are mostly holding steady from week to week, with small dips this month. So there's still time to earn interest toward that financial goal you've set for this year, but don't wait to maximize these favorable savings APYs. Here's a closer look at where rates stand and the banks that offer them. Today's best savings rates Bank APY*Min. deposit to openVaro 5.00%**$0Newtek Bank 4.55%$0LendingClub 4.50%$0Bask Bank 4.50%$0EverBank 4.30%$0Laurel Road 4.15%$0Synchrony Bank 4.10%$0American Express 3.80%$0Capital One 3.80%$0 Experts recommend comparing rates before opening a savings account to get the best APY possible. Enter your information below to get CNET's partners' best rate for your area.The savings rates average from week to week Last week's CNET average savings APY* This week's CNET average savings APYWeekly change4.20% 4.18%-0.48% Savings rates could continue to hold steadyExperts predict that the Fed will hold rates steady at this month's Federal Open Market Committee meeting, and banks are likely to follow suit. If so, you'll have more time to earn interest on your savings with over 4% APY. But even if rates drop, a high-yield savings account is still worth having."People may hesitate if it's worth it to open a HYSA with the lower rates we are currently seeing," said Danielle Flores, a CNET Money Expert Review Board member and founder of I Like to Dabble. "It is always worth it to earn a little extra on your already saved money."Savings rates are variable and will continue to fluctuate but earned interest adds up in the long run, Flores pointed out. Plus, rates for HYSAs will still be higher than most traditional savings accounts, so you'll earn more interest on your money.For example, let's say you make a one-time deposit of $500 into a HYSA with a 4.2% APY. Assuming the rate stays the same for the next 12 months, you'll earn $21.60 in interest. If you keep your money in a traditional savings account that's offering 0.42% for the same deposit you'll earn $3.60 over the same time frame. What to consider when opening a high-yield savings accountSince credit unions and online-only banks are most likely to offer high-yield savings accounts, that may change how you manage your savings. For example, some online-only banks don't accept cash deposits and may not have physical locations for in-person assistance. Here are some other factors to consider when opening a HYSA.Minimum deposit requirements: Some HYSAs require a minimum amount to open an account, typically between $25 and $100. Others don't require anything.ATM access: Not every bank offers cash deposits and withdrawals. If you need regular ATM access, check to see if your bank offers ATM fee reimbursements or a wide range of in-network ATMs, said Lanesha Mohip, founder of the Polished CFO and another CNET Expert Review Board member.Fees: Look out for fees for monthly maintenance, withdrawals and paper statements, said Mohip. The charges can eat into your balance.Accessibility: If you prefer in-person assistance, look for a bank with physical branches. If you're comfortable managing your money digitally, consider an online bank.Withdrawal limits: Some banks charge an excess withdrawal fee if you make more than six monthly withdrawals. If you may need to make more, consider a bank without this limit.Federal deposit insurance: Make sure your bank or credit union is insured with the Federal Deposit Insurance Corporation or the National Credit Union Administration, respectively. This way, your money is protected up to $250,000 per account holder, per category, if the bank fails.Customer service: Choose a bank that's responsive and makes it easy to get help with your account if you need it. Read online customer reviews and contact the bank's customer service to get a feel for working with the bank.MethodologyCNET reviewed savings accounts at more than 50 traditional and online banks, credit unions and financial institutions with nationwide services. Each account received a score between one (lowest) and five (highest). The savings accounts listed here are all insured up to $250,000 per person, per account category, per institution, by the FDIC or NCUA.CNET evaluates the best savings accounts using a set of established criteria that compares annual percentage yields, monthly fees, minimum deposits or balances, and access to physical branches. None of the banks on our list charge monthly maintenance fees. An account will rank higher for offering any of the following perks.Account bonusesAutomated savings featuresWealth management consulting/coaching servicesCash depositsExtensive ATM networks and/or ATM rebates for out-of-network ATM useA savings account may be rated lower if it doesn't have an easy-to-navigate website or if it doesn't offer helpful features like an ATM card. Accounts that impose restrictive residency requirements or fees for exceeding monthly transaction limits may also be rated lower.*APYs as of Jan. 17, 2025, based on the banks we track at CNET. Weekly percentage increase/decrease from Jan. 6, 2025, to Jan. 13, 2025.**Varo offers 5% APY only on balances of less than $5,000.More savings advice