Agrégateur de flux

HP has revealed one of the more unusual PC form factors at CES 2026: a complete Windows PC built directly into a keyboard. The device is called the EliteBoard G1a Next Gen AI PC, and unlike novelty designs of the past, it is positioned as a serious enterprise workstation rather than an experiment.

The EliteBoard G1a looks like a standard low-profile office keyboard. Inside, it houses an AMD Ryzen AI 300 Series processor, system memory, storage, connectivity, and optional battery hardware. There is no external box, no dock, and no separate compute unit. The keyboard is the computer.

A keyboard that is the entire PC

HP is describing the EliteBoard G1a as the first "AI PC built into a keyboard." The entire system measures roughly 12 mm thick and weighs between 1.5 and 1.7 pounds, depending on configuration. From the outside, it resembles a typical enterprise keyboard with chiclet-style keys, a full function row, and a numeric keypad.

This is not a thin client or a Raspberry Pi-style experiment. The EliteBoard G1a runs full Windows and is powered by modern x86 hardware. Configurations support up to 64 GB of RAM and up to 2 TB of storage, making it suitable for everyday productivity workloads, web-based tools, and standard business applications.

HP is not marketing it as a gaming machine, but it is far more capable than previous "computer-in-a-keyboard" designs that relied on low-power ARM or single-board computers.

Built around Ryzen AI and local processing

At the core of the EliteBoard G1a is an AMD Ryzen AI processor with a dedicated NPU capable of more than 50 TOPS. That places it squarely in the Copilot+ PC category, with hardware designed for local AI workloads.

HP pairs this with its own power and thermal management systems, including Auto State Management and Smart Sense. These adjust performance, cooling behavior, and power draw dynamically based on workload. For users who choose the battery-equipped version, this helps balance responsiveness with limited untethered runtime.

The optional battery allows the keyboard PC to be moved between desks or meeting rooms without immediately reconnecting power. HP also offers a non-battery configuration for fixed desks, which keeps weight down and simplifies deployment.

Designed for enterprise environments

The EliteBoard G1a is aimed squarely at business and enterprise customers. That shows up in the feature set. There is a Kensington lock slot, optional fingerprint authentication, and additional security software baked into the platform.

Connectivity is also conservative by design. Display output relies on DisplayPort rather than HDMI, reflecting typical enterprise monitor deployments. The system can drive up to two 4K displays simultaneously, which aligns with modern office workflows.

HP has confirmed that the device will be sold directly through HP.com rather than through retail channels, reinforcing that this is not a mass-market consumer product.

A clean-desk alternative to mini PCs and laptops

The appeal of the EliteBoard G1a is not raw performance. It is simplicity. For users who already work with external monitors, docks, and peripherals, a keyboard PC removes an entire layer of hardware.

There is no tower under the desk and no mini PC mounted behind a monitor. One cable handles power, display output, and data. Dual microphones and built-in speakers are integrated into the keyboard itself, allowing it to function immediately in video calls without extra peripherals.

The form factor also makes hot-desking and shared workspaces easier to manage. IT departments can deploy identical keyboards across desks, with the computer moving with the user rather than staying fixed in place.

Paired with a new IPS Black 4K monitor

Alongside the EliteBoard G1a, HP introduced the Series 7 Pro 4K Monitor as a companion display. It uses IPS Black panel technology with a claimed 2,700:1 contrast ratio, roughly double that of standard IPS displays.

The monitor is factory calibrated and supports user-defined color profiles when paired with external calibration tools. It also includes a 140W Thunderbolt 4 port capable of handling power, display output, and high-speed data over a single cable.

In practical terms, this allows the EliteBoard G1a to connect to the monitor with one cable for both power and video, turning the display into a hub for the entire setup.

Not a new idea, but a modern execution

Keyboards with built-in computers are not new. Early home computers like the Commodore 64 and Apple II placed the system board under the keys. More recently, the Raspberry Pi 400 revived the idea with a Linux-based design.

What makes the EliteBoard G1a different is that it runs modern Windows on current x86 hardware, with enterprise-grade security and support. This is not a nostalgia play or a hobbyist device. It is meant to replace small form factor desktops in controlled environments.

Availability and pricing

HP says the EliteBoard G1a Next Gen AI PC and the Series 7 Pro 4K Monitor are expected to become available on HP.com in March. Pricing has not been announced.

Whether this form factor expands beyond enterprise will depend on adoption. For now, HP is treating it as a targeted solution for organizations that want fewer boxes, cleaner desks, and centralized hardware management.

Thank you for being a Ghacks reader. The post HP Built a Full Windows PC Into a Keyboard Using Ryzen AI Hardware appeared first on gHacks Technology News.

Vous utilisez Tailscale pour créer votre petit réseau privé virtuel sans vous prendre la tête ? Moi aussi, j'adore ce truc. Mais entre nous, est-ce que vous avez déjà vérifié si votre config était vraiment sécurisée ? Genre, vraiment ?

Hé bien c'est exactement pour ça qu'un développeur a créé Tailsnitch . L'idée est simple puisqu'il s'agit de faire passer votre réseau Tailscale au peigne fin pour détecter les failles de sécurité potentielles. Et y'a de quoi faire puisque l'outil embarque 52 vérifications réparties dans 7 catégories (accès, authentification, réseau, SSH, logs, appareils et DNS).

Du coup, comment ça marche ? Vous installez le bazar avec Go :

go install github.com/Adversis/tailsnitch@latest

Ensuite, vous configurez vos credentials Tailscale (soit via OAuth avec TS_OAUTH_CLIENT_ID et TS_OAUTH_CLIENT_SECRET, soit avec une simple clé API TSKEY), et hop, vous lancez tailsnitch. En quelques secondes, l'outil analyse votre configuration et vous sort un rapport avec des niveaux de sévérité : critique, élevé, moyen, faible ou informatif.

Et là, le truc vraiment cool c'est que si vous êtes flemmard (comme moi), y'a un mode --fix interactif. L'outil vous propose de corriger les problèmes un par un. Vous validez, il corrige. Et si vraiment vous n'avez pas envie de réfléchir, y'a même un mode --auto (combiné avec --fix) qui corrige tout ce qu'il peut sans rien demander. Perso, je vous conseille quand même de lancer un --dry-run avant pour voir ce qu'il compte faire, histoire de pas avoir de mauvaises surprises.

Pour les entreprises qui doivent prouver leur conformité, Tailsnitch peut aussi exporter ses résultats au format SOC 2. De quoi rassurer vos auditeurs sans vous galérer à compiler des preuves à la main.

Parmi les trucs vérifiés, on trouve : les clés d'authentification qui n'expirent jamais (mauvais plan), les ACL trop permissives, le MagicDNS désactivé, les appareils non mis à jour, ou encore les nœuds avec le transfert SSH activé alors qu'ils ne devraient pas. Bref, tout ce qu'on oublie de vérifier quand on configure son réseau mesh à l'arrache.

L'outil est open source (licence MIT), codé en Go, et tourne sur n'importe quel OS. C'est donc un truc à lancer de temps en temps pour s'assurer que personne n'a ouvert une brèche dans votre joli tunnel chiffré.

On avait les Mindstorms, les Boost, les Powered Up, et maintenant voici la Smart Brick ! LEGO vient en effet de lâcher une bombe au CES 2026 : un micro-ordinateur complet qui tient dans une brique 2x4 classique. Et visiblement, c'est pas qu'un gadget marketing !

La Smart Brick, une brique 2x4 qui cache une puce ASIC de 4,1mm ( Source )

Le concept est fou puisque LEGO a réussi à caser une puce ASIC de 4,1mm (plus petite qu'un tenon LEGO !), des capteurs (accéléromètre, lumière, son), un mini haut-parleur avec synthétiseur intégré, du Bluetooth et des bobines de charge sans fil... le tout dans le format exact d'une brique 2x4 standard. De l'extérieur, impossible donc de faire la différence avec une brique normale.

Le truc cool, c'est que ça fonctionne avec des Smart Tags, des tuiles 2x2 qui servent à donner du contexte à la brique. Vous approchez une minifig Star Wars équipée d'un tag NFC et hop, la brique sait qu'elle doit jouer le thème de Dark Vador. Vous posez votre X-wing sur un socle taggé et les bruitages de moteur se déclenchent. Pas de caméra, pas d'IA générative, juste des interactions physiques bien pensées.

D'ailleurs, la fascination de LEGO pour l'automatisation n'est pas nouvelle. Mais cette fois, c'est directement dans les briques que ça se passe.

Le pad de recharge sans fil permet de charger jusqu'à 10 briques simultanément ( Source )

Pour la batterie, LEGO annonce plusieurs années d'autonomie en veille. Et pour la recharge, un pad sans fil à 39,99$ permet de charger jusqu'à 10 briques simultanément. Fini les piles à changer ou les câbles qui traînent partout.

Côté sets, LEGO démarre fort avec Star Wars dès le 1er mars 2026 :

• Le TIE Fighter de Dark Vador (75421) à 69,99$
• Le X-wing de Luke (75423) à 99,99$
• Le Duel de la Salle du Trône avec A-wing (75427) à 159,99$

Le TIE Fighter de Dark Vador (75421) avec Smart Bricks intégrées ( Source )

Chaque set embarque des Smart Bricks, des minifigs NFC et des accessoires son/lumière. Julia Goldin, la directrice produit et marketing de LEGO, parle carrément de "plus grande évolution depuis la minifig en 1978". Bref, ils y croient fort.

Et le meilleur dans tout ça ? La Smart Brick reste 100% compatible avec vos millions de briques existantes. Votre vieux château fort des années 90 pourra accueillir des briques qui font du bruit. De quoi verser une petite larme en retrouvant des sensations qu'on pensait à jamais disparues.

Les précommandes ouvrent le 9 janvier 2026. Après Star Wars, LEGO City, Technic et Harry Potter suivront dans l'année.

Alors, prêts à ressortir vos LEGO du placard pour y ajouter un peu de jus de cervelle ?

Source

Un jeu de course Star Wars sorti en 2002 vaut aujourd'hui plus de 400 dollars sur eBay. Pas parce qu'il est devenu culte... mais parce qu'il permet de déverrouiller les PS5.

Vous allez voir, l'histoire est dingue ! Star Wars Racer Revenge , un titre obscur de l'ère PS2, cachait depuis plus de 20 ans une vulnérabilité dans son code. Et cette faille vient d'être exploitée pour extraire les clés ROM de la PlayStation 5 , c'est à dire le Saint-Graal des hackers de consoles dont je vous parlais y'a 5 jours...

L'édition PS4 de Limited Run Games - 8 500 exemplaires seulement ( Source )

Le chercheur en sécurité Gezine a balancé la nouvelle sur Twitter le 31 décembre dernier, du coup, le prix des copies physiques de la version PS4 (éditée par Limited Run Games en 2019 à seulement 8 500 exemplaires) est passé de 20 dollars à plus de 400 dollars en quelques heures. Une hausse de 1900% ! Y'a des gens qui ont fait une sacrée affaire ce jour-là...

L'exploit utilise la technique " mast1core " et cible le firmware 12.00 de la console. En gros, le menu "Hall of Fame" du jeu contient une faille qui permet d'injecter du code et comme Sony n'a jamais pensé à patcher un vieux jeu de podracing de l'époque de la préquelle, eh bien... la porte est restée grande ouverte pendant deux décennies.

Le truc vraiment flippant pour Sony, c'est que les clés bootrom extraites sont au niveau le plus fondamental du système. On ne parle pas d'une faille logicielle qu'une mise à jour peut corriger. Non, ces clés sont gravées pour toujours dans le silicium, donc c'est impossible à patcher ! Sony peut toujours sortir des firmwares pour compliquer la tâche, mais le mal est fait.

Pour ceux qui espéraient chopper une copie, c'est un peu tard par contre. Les exemplaires physiques s'arrachent sur eBay entre 230 et 500 dollars. Perso, j'aurais jamais imaginé qu'un jeu Star Wars médiocre deviendrait un jour un outil de hacking aussi précieux. D'ailleurs, si vous l'avez chez vous, c'est peut-être le moment de le revendre ^^.

Et le plus marrant dans tout ça c'est Limited Run Games avait produit ce jeu en petite quantité justement parce que personne n'en voulait vraiment.

Et dire qu'aujourd'hui, chaque copie est devenue une clé magique pour faire tourner du homebrew sur la dernière console de Sony...

Source

Vous vous souvenez du dark web , ce truc mystérieux qui fait flipper les politiques, les journalistes et votre belle-maman avec Tor, les .onion et tout le folklore ? Hé bien figurez-vous qu'une bonne partie du crime en ligne s'est déplacée ailleurs. Maintenant en 2026, pour monter le plus grand marché illicite en ligne jamais mesuré, y'a plus besoin de toute cette sophistication technique.

Y'a juste à aller sur Telegram, une simple app de messagerie, et hop, c'est parti.

La firme d'analyse crypto Elliptic vient en effet de publier des chiffres qui donnent le tournis. Huione Guarantee, une plateforme de "garantie" pour arnaqueurs sinophones hébergée sur Telegram, a brassé 27 milliards de dollars de transactions entre 2021 et 2025. Vingt-sept milliards les gars !

Pour vous donner une idée, AlphaBay , le célèbre marché noir du dark web démantelé par les autorités en 2017 et considéré comme dix fois plus gros que Silk Road , n'avait fait "que" 1 milliard en un peu plus de deux ans. Hydra, le mastodonte russe qui a terrorisé les autorités pendant environ six ans, culminait quand à lui à 5 milliards.

Hé bien Huione les a atomisés en opérant... en plein jour sur une app grand public.

Chute brutale des volumes Huione après le ban Telegram du 13 mai 2025 ( Source Elliptic )

Finalement, le 13 mai 2025, Telegram a fini par bannir Huione Guarantee. Victoire ? Pas vraiment. Car deux marchés concurrents ont immédiatement pris le relais : Tudou Guarantee et Xinbi Guarantee.

Et là, le magie-magie, on découvre que Huione détient 30% des parts de Tudou. Autant dire qu'ils ont juste changé d'enseigne...

Et aujourd'hui, ces deux plateformes traitent environ 2 milliards de dollars par mois en blanchiment d'argent, vente de données volées, faux sites d'investissement, outils de deepfake et autres joyeusetés. Elliptic surveille actuellement plus de 30 marchés de ce type sur Telegram et c'est pas triste...

Tudou prend le relais : explosion des volumes après la fermeture de Huione ( Source Elliptic )

Et tout ça sert à quoi ?

Principalement aux arnaques dites "pig butchering" (littéralement "boucherie de cochon"), ces escroqueries sentimentalo-crypto où l'arnaqueur vous "engraisse" patiemment avant de vous plumer. Le FBI estime que ces arnaques siphonnent près de 10 milliards de dollars par an aux victimes américaines.

Juste aux États-Unis, hein. Mais nous aussi on en reçoit tous les jours...

Et derrière ces arnaques, on trouve malheureusement des centres au Cambodge, au Myanmar et au Laos où des milliers de personnes sont forcées de travailler sous la menace. C'est ça l'envers cauchemardesque de ces chiffres astronomiques.

Ce qui me sidère, c'est que pendant des années, on nous a vendu le dark web comme l'eldorado intraçable des criminels, avec Tor, le chiffrement bout en bout, les protocoles anonymes... Et au final, le plus grand marché illicite de tous les temps opère sur une app que votre grand-mère pourrait installer. Les seules compétences requises c'est juste de parler chinois et d'être assez teigneux pour recréer un channel quand on se fait bannir.

Tom Robinson, cofondateur d'Elliptic, explique que c'est tout simplement "le plus grand marché illicite en ligne à avoir jamais existé". Plus grand que Silk Road, AlphaBay et Hydra réunis, tout ça OKLM sur Telegram.

On se demande ce que fout Pavel Durov... hmm...

En tout cas, ça montre que la sophistication technique n'est pas le facteur limitant du crime en ligne. Par contre, on est visiblement loin d'avoir résolu le problème. Quand Telegram en bannit un, y'en a deux autres qui poussent. C'est sans fin !

Bref, la prochaine fois qu'on vous parle du "dark web" comme de la grande menace, vous pourrez sourire car le vrai game se joue sur des apps qu'on trouve en libre accès sur l'App Store.

Source

Le salon ITB Berlin, dédié au marché touristique mondial se déroulera du 03 au 05 mars 2026.
A lire ou relire :
Livre blanc : Boostez les revenus de votre groupe hôtelier avec Mews
Dans son livre blanc, Mews a identifié 8 méthodes pour accompagner les groupes hôteliers dans leur développement et leur prospérité.
L'avis de Guilain Denisselle : Gérer un groupe hôtelier demande (...)

Renseignements & inscription : (...)

- Agenda / ITB, Berlin, salon, conférence, tourisme

À la suite de l'incendie meurtrier survenu lors du réveillon dans un lounge bar en Suisse, ayant causé plusieurs dizaines de morts et plus d'une centaine de blessés, les professionnels réunis au sein du Groupement des Hôtelleries & Restaurations de France (GHR) souhaitent exprimer leur profonde tristesse et leur soutien le plus sincère aux familles des victimes, aux personnes blessées, ainsi qu'à l'ensemble de celles et ceux durablement marqués par ce drame.
Les pensées des professionnels vont (...)

- Communiqué de presse / GHR, Crans-Montana, actualité

Food Hotel Tech présente toutes les innovations dédiées aux professionnels de l'hôtellerie et de la restauration afin de les aider à accélérer leur transformation durable et digitale qui leur permettra de satisfaire les nouvelles attentes des clients, du personnel et d'améliorer leur marge.
Exposants, village startups, conférences et networking : découvrez à FHT les meilleures innovations, profitez d'un panorama complet des outils digitaux et éco-responsables et bénéficiez de conseils d'experts et de (...)

- Agenda / Food Hotel Tech, Paris, hoteltech, technologie, FoodTech, conférence, salon, innovation

Le Salon IODE 2026 : l'Événement Incontournable pour tous les professionnels de l'hôtellerie de plein air et traditionnelle les 28 & 29 janvier 2026 à Vannes !
L'industrie du tourisme connaît une dynamique constante, et pour demeurer à la pointe des dernières tendances et innovations, les gérants de campings, hôtels, parcs de loisirs, collectivités et autres acteurs du secteur de l'hôtellerie & du tourisme ont rendez-vous avec un des événements phares de l'année : le Salon IODE. Cet événement (...)

- Agenda / IODE, Vannes , salon

Cette année, le Grand Live continuera de donner la parole à celles et ceux qui font bouger notre secteur : travel managers, acheteurs, hôteliers, TMC, experts et acteurs engagés de l'écosystème.
Tendances, innovations, durabilité, technologie, expérience voyageur… Autant de sujets au cœur des enjeux d'aujourd'hui et de demain à découvrir en live. Quelle économie en 2026 ? Marc Touati, économiste de renom, Président du cabinet ACDEFI et Conseiller économique d'eToro, dressera un premier bilan de la (...)

- Agenda / CDS, voyage d'affaires, conférence

Samsung has revealed its largest and most ambitious Micro RGB television yet at CES 2026. The new 130-inch model, known as the R95H, is positioned as a statement product that blends display technology, industrial design, and AI-driven features into a single showcase.

The size alone puts the TV well outside mainstream living rooms, but Samsung is using it to highlight where its display roadmap is heading rather than to chase volume sales.

Micro RGB pushes color beyond existing TV tech

Micro RGB is the core of the announcement. Unlike traditional LCD TVs that rely on a backlight, or OLED panels that use organic compounds, Micro RGB uses microscopic red, green, and blue LEDs that emit light independently. Samsung says this allows the R95H to cover 100% of the BT.2020 color gamut, which is wider than what current consumer OLED and Mini-LED TVs can deliver.

Samsung pairs the panel with its new Micro RGB AI Engine Pro, along with Color Booster Pro and HDR Pro processing. These systems are designed to preserve color accuracy while boosting contrast and shadow detail, even at extreme brightness levels. A glare-free coating is also used to maintain image clarity in brightly lit rooms.

A TV designed to look like part of the room

Samsung is leaning heavily into design with what it calls the "Timeless Frame." Instead of looking like a conventional television, the screen is meant to resemble a floating gallery window. The frame hides integrated spatial audio hardware that is tuned specifically for the size of the display, allowing sound to track on-screen action without external speakers.

This approach mirrors Samsung's ongoing effort to make large TVs feel more architectural and less like standalone electronics.

Vision AI Companion goes beyond voice commands

The R95H ships with Samsung's updated Vision AI Companion, an on-device assistant that extends past basic voice control. In addition to content recommendations, it can answer contextual questions about what's on screen, suggest recipes, and offer music or program ideas based on viewing habits.

Samsung is also introducing AI Soccer Mode Pro, which adjusts picture and sound settings to create a stadium-style presentation, and an AI Sound Controller Pro that lets viewers separately tune commentary, crowd noise, and background audio.

Long-term software support and broader 2026 lineup

At this price tier, Samsung is bundling its latest software platform as well. The TV supports HDR10+ Advanced, the company's Eclipsa spatial audio system, and a Tizen-based interface now promised to receive seven years of updates.

The 130-inch Micro RGB TV sits alongside Samsung's refreshed 2026 lineup, which includes a slimmer OLED S95H and a new Freestyle+ portable projector. Those products are expected to be far more accessible, but the Micro RGB model is clearly the centerpiece of Samsung's CES presence.

Samsung has not announced pricing or availability for the R95H or the rest of its 2026 TV lineup.

Thank you for being a Ghacks reader. The post Samsung Unveils a 130-inch Micro RGB TV With Gallery-Style Design and New AI features appeared first on gHacks Technology News.

Asus may be quietly winding down its smartphone ambitions. According to distribution channels in Taiwan, the company has decided not to launch any new smartphones in 2026, putting its long-running gaming-focused ROG Phone lineup on pause.

For a brand that never chased volume but built its reputation on extreme performance, the move suggests the niche may no longer make sense commercially.

No new Asus phones planned for 2026

Distributors in Taiwan have reportedly been unable to source new Asus smartphones for some time. Information shared with partners indicates that Asus continued operating its smartphone business only through the end of 2025.

Asus has since confirmed that it has informed telecom partners of its plan not to release new smartphones in 2026. The company says it will continue supporting existing devices with software updates, warranty coverage, and after-sales service.

What that means in practical terms is simple: there will be no ROG Phone 10, and no new Zenfone flagship this year.

The statement stops short of announcing a permanent exit. Asus frames the move as a pause rather than a shutdown, leaving the door open to a possible return. Still, the absence of any forward-looking roadmap makes the future of its phone business uncertain.

A niche that may no longer pay off

The ROG Phone carved out a distinct identity in the Android space. Large batteries, aggressive cooling, shoulder triggers, and maximum chipset performance made it a favorite among hardcore mobile gamers.

That focus also limited its reach. ROG Phones were expensive, bulky, and often hard to find outside select markets. As mainstream Android flagships closed the performance gap and mobile gaming shifted toward efficiency rather than raw power, the value proposition narrowed.

From a business perspective, maintaining a smartphone division for a small, performance-obsessed audience may no longer justify the cost.

Why Asus might be pulling back

Asus has not given a detailed explanation for the decision. Several pressures line up, though.

Component costs remain volatile, particularly DRAM and NAND pricing. Smartphones are also facing slowing global demand, tighter margins, and increased competition from Chinese brands that operate at massive scale.

Gaming phones sit at the intersection of all of these problems. They rely on premium components, sell in limited volumes, and require ongoing software support to stay relevant.

Main changes for users

For most smartphone buyers, the impact is minimal. Asus was never a major player in the broader Android market. Its Zenfone and ROG devices appealed to enthusiasts, not the mass market.

For fans of gaming phones, however, this removes one of the few brands willing to build devices around performance first, battery size second, and everything else after that.

Asus says existing devices will continue receiving updates and support, so current owners are not being abandoned. The bigger question is whether the pause becomes permanent.

If Asus does return, it will likely do so in a very different market than the one that allowed gaming phones to stand out in the first place.

Thank you for being a Ghacks reader. The post Asus Appears To Be Stepping Away From Gaming Phones in 2026 appeared first on gHacks Technology News.

Boston Dynamics is moving its humanoid robot Atlas out of controlled demos and into real industrial environments. The company is now testing AI-powered versions of Atlas inside factories, with early trials focused on repetitive material-handling work.

With backing from Hyundai, which owns a majority stake in Boston Dynamics, Atlas has already been deployed in limited trials at Hyundai's new Georgia manufacturing plant. There, the humanoid robot practiced autonomously sorting roof racks for the assembly line, a task normally handled by human workers.

This marks a shift from spectacle to practicality. Atlas is no longer just running, jumping, or performing choreographed routines. It is being trained to do useful work.

From pre-programmed moves to learned behavior

Earlier versions of Atlas relied heavily on manually coded motion algorithms. The current generation looks very different. It is fully electric, slimmer, and powered by AI systems running on high-end compute hardware.

Instead of scripting every movement, Boston Dynamics now trains Atlas using machine learning. Engineers teach the robot tasks through demonstrations rather than explicit instructions.

One method involves supervised learning using virtual reality. A human operator wears a VR headset and directly controls Atlas' arms and hands, guiding it through each step of a task. That interaction generates training data, which is then used to teach Atlas how to repeat the task autonomously.

Another technique uses motion capture suits. Human movements are recorded and translated to Atlas' very different body structure. The robot does not copy motions directly but learns how to achieve the same outcome using its own mechanics.

Training thousands of robots at once

Much of Atlas' learning happens in simulation. Boston Dynamics runs thousands of digital Atlas models in parallel, exposing them to different conditions such as uneven floors, limited mobility, or mechanical resistance.

These simulations allow the system to test variations quickly and identify stable movement strategies. Once a skill is learned in simulation, it is uploaded to the control system used by every physical Atlas robot.

That approach means improvements scale instantly. Training one Atlas effectively trains them all.

Useful, but far from general-purpose

Atlas can now run, crawl, lift, sort objects, and perform coordinated movements that were considered unrealistic for humanoid robots only a few years ago. Still, Boston Dynamics is clear about the limitations.

The robot cannot yet perform many everyday tasks humans take for granted, such as dressing, cooking, or handling fragile objects reliably. These remain difficult problems due to the complexity of human environments and fine motor control.

The company's focus is narrower: repetitive, physically demanding work in structured environments like factories and warehouses.

Factory work comes first

The Hyundai trial highlights where humanoid robots are most likely to appear first. Assembly-line logistics, material handling, and physically taxing jobs are well-defined and repeatable, making them ideal early use cases.

Boston Dynamics' leadership has framed this as a way to reduce human exposure to dangerous or exhausting work rather than replace entire roles. Robots like Atlas still require human oversight, maintenance, and training.

That also limits how autonomous they can be in the near term.

No Terminator scenario

Despite growing concern about AI and automation, Boston Dynamics downplays fears of runaway humanoids. The company describes the reality as the opposite: robots are difficult to train, fragile in unexpected situations, and heavily dependent on human input.

Atlas represents progress, but also the current ceiling of what AI-powered humanoids can do outside controlled environments.

For now, the goal is not general-purpose robots roaming freely, but specialized machines capable of doing one job well and safely.

Atlas' factory trials suggest that phase is already underway.

Do you believe that robots can replace humans in manual work? yes/no

Thank you for being a Ghacks reader. The post Boston Dynamics Is Training Atlas to Handle Real Factory Tasks With AI appeared first on gHacks Technology News.

When hardware behaves oddly on a Linux system, the fastest answers are often already available from the terminal. One of the most useful tools for this is lspci, a small command that reports what devices are connected to the PCI bus and how the kernel sees them.

lspci is not flashy, but it is reliable. It works on servers, laptops, desktops, and recovery environments, and it does not depend on a desktop session being available. These are the ways it is most useful in day-to-day troubleshooting.

1. Get a quick inventory of internal hardware

Running lspci without any flags gives a concise list of PCI devices detected by the system:

lspci

This shows graphics adapters, network controllers, storage controllers, and chipset components. Each entry includes a slot address, vendor name, and device description. If a GPU, network card, or controller does not appear here, Linux is not seeing it at all.

This step is often enough to confirm whether a hardware issue is physical or software-related.

2. Narrow the output to a specific device

On systems with lots of devices, the output can be overwhelming. Piping the results into grep makes it easier to focus on one component:

lspci | grep USB

This works well for USB controllers, graphics devices, or network hardware. Searching by vendor name is often more reliable than generic terms like "GPU" or "graphics," which may not appear in the device label.

3. View detailed information for troubleshooting

When basic output is not enough, verbose mode adds useful context:

lspci -v

This includes IRQ assignments, memory ranges, and other technical details that help diagnose driver or resource conflicts. Some fields require elevated privileges, so running it with sudo may reveal more data.

Higher verbosity levels exist, but they are usually unnecessary unless debugging kernel or driver-level issues.

4. Understand how devices are connected

The PCI tree view shows how devices relate to each other:

lspci -tv

This is especially useful on laptops and servers, where multiple devices share bridges or controllers. It can explain why disabling or removing one device affects another, and it helps when diagnosing bandwidth or bus contention problems.

5. Check which driver a device is using

Knowing whether the correct kernel driver is loaded can save time. Once you know the device's slot address, you can check its driver like this:

lspci -ks 00:02.0

This reveals the kernel module currently in use and which modules are available. For graphics cards and network adapters, this step quickly shows whether the system is using an open-source driver, a fallback module, or nothing at all.

6. Get vendor and device IDs for deeper research

Sometimes device names are not enough. lspci can display numeric vendor and device IDs that uniquely identify hardware:

lspci -nn

To focus on a single device:

lspci -nns 00:02.0

These IDs are invaluable when searching documentation, reporting bugs, or checking compatibility lists. They also help confirm the exact hardware variant when multiple revisions exist.

When lspci is the right tool

lspci only shows PCI devices. It does not list USB peripherals, storage partitions, or removable drives. For those, tools like lsusb and lsblk are more appropriate.

Still, for diagnosing GPUs, Wi-Fi cards, Ethernet controllers, NVMe interfaces, and chipset components, lspci is one of the fastest ways to see what Linux actually recognizes.

Keeping it in your troubleshooting toolkit makes hardware issues easier to identify before they turn into guesswork.

 

Resources:

https://www.howtogeek.com/practical-uses-for-the-linux-lspci-command/

Thank you for being a Ghacks reader. The post Six Practical Ways to Use lspci When Troubleshooting Linux Hardware appeared first on gHacks Technology News.

CES 2026 has not officially opened yet, but the pattern is already clear. As usual, the real show starts early, with companies teasing products and half-announcements well before the doors open in Las Vegas. This year feels busier than the last few editions, with more hardware, bigger booths, and less restraint around buzzwords.

Artificial intelligence is going to be everywhere. Not as a category, but as a label glued onto almost every product. Laptops, TVs, appliances, wearables, cars-if it has a chip, it will claim some form of AI advantage. Much of it will be backend automation rather than visible features, but marketing will make sure the word is front and center.

Smart glasses look set to dominate the show floor. After years of false starts, nearly every major brand now wants a piece of what comes after the smartphone. Expect a wide mix of designs: camera-focused glasses, audio-first "AI glasses," lightweight display glasses, and hybrids that sit somewhere between eyewear and XR headsets. No single design is likely to emerge as the obvious winner, but the sheer volume will be hard to miss.

TV technology is also back in the spotlight. Display makers will push new panel types, higher brightness, and higher refresh rates, whether or not those gains translate into real-world improvements. AI processing will be heavily promoted here as well, often tied to image enhancement, motion smoothing, and content upscaling.

Mobility continues to expand its footprint at CES. Electric vehicles, e-bikes, scooters, and concept transport devices will take up large sections of the show. One noticeable shift is the slow return of physical controls in car interiors, after years of touchscreen-heavy designs. Buttons and dials are quietly becoming a selling point again.

Robotics will blur further into the smart home category. Beyond robot vacuums, expect more humanoid and semi-humanoid demos focused on lifting, carrying, and basic household tasks. These systems are not close to mass-market pricing, but CES remains a favorite venue for showing what might arrive several product cycles from now.

Alongside all of that will be the usual CES staples: laptops, monitors, audio gear, accessories, and experimental form factors that may or may not ever ship. The mix is familiar, but the density feels higher this year.

CES 2026 looks less like a reset and more like a return to excess, with AI as the connective tissue tying together a very crowded hardware show floor.

Does someone heading to CES this year?

Thank you for being a Ghacks reader. The post CES 2026 Is Where Every Tech Trend Shows Up At Once appeared first on gHacks Technology News.

Vous avez déjà écrit un programme qui affiche ces deux mots magiques ?

HELLO WORLD

Évidemment que oui. Tout le monde est passé par là car c'est le rite initiatique universel de la programmation, le premier truc qu'on tape quand on découvre un nouveau langage.

Mais est-ce que vous vous êtes déjà demandé d'où venait cette tradition bizarre de saluer le monde avant de faire quoi que ce soit d'utile ?

Hé bien ça remonte au début des années 70, aux Bell Labs. Brian Kernighan, chercheur canadien qui bossait aux côtés de Dennis Ritchie (le créateur du C) et Ken Thompson, devait rédiger un tutoriel pour le langage B. Le document s'appelait "A Tutorial Introduction to the Language B", et c'est là-dedans que la fameuse phrase apparaît pour la première fois dans un document technique.

Le Hello World original de 1978, imprimé sur papier d'imprimante matricielle ( Source )

Alors pourquoi ces mots-là et pas autre chose ?

Hé bien Kernighan lui-même a raconté l'anecdote dans une interview pour Forbes India. Il avoue que sa mémoire est un peu floue, mais il se souvient d'un dessin animé avec un poussin qui sortait de son œuf en lançant au monde ce premier message. L'image lui était restée en tête et quand il a dû trouver un exemple à afficher, c'est sorti tout seul.

Marrant non, qu'un des rituels les plus universels de l'informatique mondiale vienne d'un gag de dessin animé avec un poussin. C'est peut-être Calimero, qui sait ?

Ce premier exemple dans le bouquin n'a pas explosé par contre... Suite à cela, il y a eu un autre tutoriel en 1974 (pour le C cette fois), mais c'est vraiment en 1978 avec la publication du livre "The C Programming Language" co-écrit par Kernighan et Ritchie (le fameux K&R que tous les vieux de la vieille connaissent) que c'est devenu LA référence absolue. Ce bouquin a tellement marqué l'histoire que son premier exemple de code est devenu une tradition planétaire.

Voilà, l'anecdote est chouette et je trouve ça génial qu'un truc aussi naze datant des années 70 soit devenu un symbole universel plus de 50 ans après, et cela peu importe le langage !

Source

On a soccer pitch outside Beijing, the players are humanoid robots. They dribble, fall over, get back up, and occasionally crash into things they should not. This is not a gimmick demo. It is how a growing number of Chinese robotics companies are training machines for real-world use.

One of them is Booster Robotics, founded in 2023 by Cheng Hao. His company builds humanoid robots designed to play soccer using artificial intelligence. The goal is not entertainment alone. Soccer is a stress test.

Robot sports have become a proving ground across China. In 2025 alone, humanoid robots danced during the Spring Festival Gala, ran half-marathons, boxed, and competed in the world's first World Humanoid Robot Games in Beijing. Soccer, boxing, sprinting, and even simulated factory tasks were all part of the program.

Sports expose weaknesses quickly. Balance fails, vision systems misjudge distance, and coordination breaks down under pressure. For robotics engineers, those failures are useful. A robot that can run, turn, react, and cooperate with teammates on a field is closer to working safely in unpredictable environments.

Soccer has long been a benchmark task in robotics research. The international RoboCup competition, launched in the 1990s, uses the sport to test motion control, vision, planning, and team coordination. Cheng's team at Booster Robotics treats it the same way. The game is the lab.

The attention helps, too. Robot sports draw crowds, livestreams, sponsors, and investors. Booster Robotics ran an exhibition robot soccer league in mid-2025 that sold hundreds of tickets and attracted national broadcast coverage. Two days after winning RoboCup 2025 in Brazil, the company announced more than $14 million in new funding.

This public-facing angle fits neatly into a broader national push. China has spent the past decade accelerating its robotics industry, backed by subsidies, research funding, and local government support. Humanoid robots are now positioned as a strategic technology, tied to productivity gains and an aging population.

The government's involvement is visible. The World Humanoid Robot Games were co-hosted by Beijing authorities and state media. Regional governments have organized robot marathons and competitions, often paired with investment showcases. For now, the ecosystem still leans heavily on public backing.

The robots themselves are not flawless. At Beijing's games, humanoids ran into referees, missed punches, and collapsed mid-match. Engineers hovered nearby, resetting systems and collecting data. That is part of the process. Each failure feeds the next iteration.

What is changing is where this leads. Companies are already moving robots out of arenas and into factories. Sorting, inspection, and material handling are common test cases. Some humanoids are being trialed in controlled service environments as well.

Booster Robotics is aiming further. Just months after a lab visit, the company launched a smaller humanoid robot designed for broader use, priced far below its earlier competition-focused models. The pitch was straightforward: robots as practical helpers, not lab curiosities.

The idea being sold is familiarity. A humanoid that walks, carries objects, follows instructions, and interacts naturally fits more easily into homes and workplaces than specialized machines. Sports are just the training montage.

The robot sports craze may look playful on the surface, but it is increasingly tied to commercial strategy. The field is crowded, competition is intense, and differentiation matters. For many Chinese startups, turning robots into athletes is a way to turn them into products.

I think robots are the future of not just efficiency, but also entertainment.

 

Resources:

https://edition.cnn.com/2026/01/02/china/china-humanoid-robot-sports-intl-hnk-dst?cid=external-feeds_iluminar_flipboard

Thank you for being a Ghacks reader. The post China’s Robot Sports Boom is Turning Humanoids Into Consumer Products appeared first on gHacks Technology News.

China has enacted a major revision of its Cybersecurity Law, effective January 1, 2026. The amendments mark the most significant shift since the law's original introduction in 2017 and materially change how companies must handle cyber incidents, regulatory reporting, and compliance exposure.

The updated framework places speed and accountability at the center of enforcement. Incident response is no longer measured in days. In several cases, regulators now expect disclosure within minutes of detection.

Incident reporting timelines shrink dramatically

The most immediate operational change is the new reporting requirement for cybersecurity incidents. Operators of critical information infrastructure, and in some cases general network operators, must notify authorities of significant incidents within extremely short windows.

Depending on severity, initial reporting is required within four hours, or as little as 60 minutes. These timelines are reinforced by the Administrative Measures for National Cybersecurity Incident Reporting, which came into force on November 1, 2025, and consolidate reporting rules under a single framework enforced by the Cyberspace Administration of China (CAC). Incidents are classified into four severity levels. "Relatively major" incidents include data breaches affecting more than one million individuals or financial losses exceeding RMB 5 million.

These must be reported within four hours of discovery, followed by a detailed assessment within 72 hours and a post-incident report within 30 days. At the highest level, "particularly serious" incidents must be reported within one hour. Authorities are then required to escalate the report to national regulators and the State Council within 30 minutes, compressing escalation timelines to an unprecedented degree.

Higher penalties and personal accountability

The amended law significantly increases penalties for non-compliance. Organizations found to be in serious violation can now face fines of up to RMB 10 million. Individuals directly responsible, including executives and security leadership, may be fined up to RMB 1 million.

Enforcement procedures have also changed. Regulators are no longer required to issue warnings or remediation orders before imposing penalties. This allows authorities to move directly to sanctions, reducing the time organizations have to correct deficiencies after an incident.

Supply chain risk is explicitly addressed as well. Operators of critical infrastructure may be penalized for using non-compliant products or services, with fines in some cases reaching up to ten times the procurement value. Vendor selection and third-party risk management now carry direct regulatory consequences.

Expanded reach beyond China's borders

The revised law broadens its extraterritorial scope. Earlier versions focused on foreign activities that directly harmed China's critical information infrastructure. The amended language extends jurisdiction to foreign conduct that endangers China's network security more broadly.

This expansion affects multinational organizations with indirect exposure, including cloud services, software dependencies, managed service providers, and manufacturing or logistics systems that intersect with China-connected networks. In severe cases, authorities are authorized to impose measures such as asset freezes or other sanctions. For global enterprises, compliance obligations can now arise from architectural and operational decisions made entirely outside China.

Artificial intelligence enters the legal framework

For the first time, artificial intelligence is explicitly referenced in the Cybersecurity Law. The amendments promote the use of AI to enhance cybersecurity management while simultaneously calling for stronger ethics oversight and safety governance.

The law does not yet define detailed AI compliance requirements. Those are expected to emerge through follow-up regulations or technical standards. The inclusion itself signals that cybersecurity compliance in China is expanding beyond traditional infrastructure security into algorithmic risk and system-level accountability.

Clear thresholds for severe incidents

The CAC's reporting measures also define what qualifies as a "particularly serious" incident. Examples include cyber incidents that disable government portals or major news platforms for more than 24 hours, or six hours in cases of complete system failure. Large-scale disruptions affecting essential services for more than half of a province's population, or impacting the daily lives of more than 10 million people, are also included.

Data breaches involving personal information of more than 100 million individuals or financial losses exceeding RMB 100 million fall into the same category. Once an incident is resolved, operators must submit a comprehensive report within 30 days covering root causes, response actions, impact, corrective measures, and lessons learned.

What organizations should be doing now

The practical impact of the amendments is immediate. Incident response plans that assume extended investigation periods no longer align with legal requirements. Security teams must be able to classify incidents, assess severity, and trigger regulatory notification almost immediately.

Decision-making authority may need to be delegated in advance, especially for multinational organizations operating across time zones. Evidence collection and documentation processes must function in parallel with response, not after containment. For companies connected to Chinese infrastructure through suppliers, software, or services, the amended law turns speed and documentation into enforceable legal obligations rather than best practices.

Thank you for being a Ghacks reader. The post China’s New Cybersecurity Law Demands Faster Incident Reporting From Companies appeared first on gHacks Technology News.

Most calendars end up as passive records of meetings. Events go in, notifications pop up, and everything else that needs doing floats around as an informal mental list. That setup works until days start feeling full without much actually getting finished.

The single Google Calendar feature that changes this dynamic is time blocking. Instead of treating the calendar as a log, it turns it into a planning tool by assigning chunks of the day to specific types of work. Time blocking is simple. Tasks are no longer abstract items on a to-do list.

They are given a start time and an end time and placed directly alongside meetings. Writing, admin work, research, exercise, and even breaks become calendar events. This solves a common problem with task-based productivity systems. A task list shows what needs doing but ignores when it will actually happen. Without time attached, everything competes for attention at once.

Time blocking forces prioritization by making the limits of the day visible. In practice, this exposes unrealistic plans quickly. If the calendar is already full, adding another block means something else has to move. That trade-off is useful. It prevents overcommitting and makes delays obvious instead of hidden. The feature works best when blocks describe intent rather than micromanaged tasks. A "writing" block or "admin work" block leaves room for flexibility while still protecting time.

If something unexpected comes up, blocks can be dragged to another slot instead of being abandoned entirely. Google Calendar keeps the process frictionless. Creating a block takes seconds, recurring blocks handle daily routines, and color coding makes different types of work easy to scan at a glance. If a block does not get used as planned, rescheduling it keeps the plan intact rather than discarding it.

Time blocking does not add hours to the day. It reduces the mental overhead of constantly deciding what to work on next. The calendar answers that question in advance. Once the habit sticks, the calendar stops being a passive reminder system and starts acting as a boundary around time. That boundary is what makes the difference.

Thank you for being a Ghacks reader. The post Time Blocking in Google Calendar is The One Feature That Keeps Daily Schedules Under Control appeared first on gHacks Technology News.

Samsung is entering CES 2026 with a broader scope than in previous years. The company is using the event to showcase new display technology, gaming hardware, audio products, foldable devices, and a unified push around AI-powered consumer electronics.

Unlike years where the focus centered almost entirely on televisions, Samsung is spreading its announcements across multiple product categories. That approach is reflected in the volume of pre-CES disclosures and the decision to operate a large standalone exhibition space away from the main show floor. CES has traditionally been one of Samsung's most important public stages, and 2026 follows that pattern, with several product lines positioned as near-term commercial offerings rather than distant concepts.

Micro RGB televisions expand beyond early adopters

Samsung is using CES 2026 to formally expand its Micro RGB television lineup. The company has confirmed models ranging from 55 inches up to 115 inches, building on a limited release strategy that began in 2025. Micro RGB is positioned as an alternative to OLED rather than a replacement.

The technology uses microscopic RGB LEDs instead of organic compounds, with the stated goal of delivering OLED-like contrast and color accuracy while improving longevity and reducing burn-in risk. Samsung is also emphasizing scalability, which explains the unusually wide size range announced for 2026. Pricing and regional availability remain unclear, but the move signals that Micro RGB is transitioning from showcase hardware into a broader consumer-facing category.

Gaming monitors push resolution and refresh rate limits

Gaming displays remain a CES staple for Samsung, and the 2026 lineup continues that trend. The company is bringing a refreshed Odyssey monitor range, with a focus on extremes rather than incremental upgrades. Among the announced models is the Odyssey 3D, a glasses-free 3D gaming monitor with a 6K resolution.

Samsung is also highlighting the Odyssey G6, which it describes as the first gaming monitor capable of reaching a 1,040Hz refresh rate. Additional Odyssey G8 models include 6K, 5K, and OLED variants. These products are aimed at niche segments-competitive gamers, enthusiasts, and early adopters-rather than mainstream users. CES will be the first opportunity to evaluate how usable these specifications are outside of controlled demos.

Foldables make an appearance beyond phones

Samsung's foldable strategy is also visible at CES 2026. The company's first tri-fold smartphone, the Galaxy TriFold, was already unveiled prior to the show, but CES provides a public hands-on opportunity for many attendees. While no additional foldable phones are expected to be formally launched during the event, Samsung Display is expected to exhibit experimental panels and form factors.

Historically, these display demos have served as early indicators of hardware that appears in consumer devices one or two product cycles later. Samsung has not confirmed timelines or commercial plans for additional foldable formats shown at CES.

Audio products emphasize design and wireless features

Samsung is introducing new audio hardware at CES 2026, including additions to its Q-series soundbar lineup. More attention is likely to fall on two new Wi-Fi speakers: the Music Studio 5 and Music Studio 7. The Music Studio 7 is a 3.1.1-channel system with front-facing, side, and top-firing drivers designed to create spatial audio effects without external speakers. It supports high-resolution audio and integrates Samsung's AI-based audio processing features.

The smaller Music Studio 5 focuses on compact design, using a four-inch woofer and dual tweeters. Both speakers support Wi-Fi streaming, Bluetooth connectivity, and Samsung's ecosystem features. Industrial design is a notable aspect, with Samsung crediting designer Erwan Bouroullec for the speakers' external appearance.

AI becomes the unifying theme

Samsung is framing CES 2026 around what it calls a unified AI approach across its Device eXperience division. To support that message, the company has set up a large standalone exhibition space at The Wynn, separate from the main convention center.

The AI focus spans multiple product categories. Samsung has confirmed new AI-connected home appliances under its Bespoke branding, including washing machines, air conditioners, clothing care systems, refrigerators, and robotic cleaners. These devices emphasize automation, usage prediction, and integration with other Samsung hardware.

Samsung is also extending its partnership with Google, integrating Gemini-based features into select home appliances. The goal is to make voice interaction and contextual automation consistent across different device types. The scale of Samsung's CES presence suggests that the exhibition itself is intended to be part of the announcement, offering hands-on demonstrations rather than relying solely on staged presentations.

What this means for users

CES 2026 shows Samsung doubling down on hardware breadth rather than narrowing its focus. Displays, gaming gear, foldables, audio products, and home appliances are all being positioned as parts of a single ecosystem, tied together by software and AI services.

For users, the practical impact will vary by category. Some products, such as Micro RGB TVs and new Odyssey monitors, are moving closer to retail availability. Others, particularly AI-driven appliances and experimental displays, are more about signaling long-term direction. Samsung has not provided release dates or pricing for most of the hardware shown at CES, and availability may vary significantly by region.

Are you a fan of Samsung? Or prefer Apple technologies?

Thank you for being a Ghacks reader. The post Samsung Outlines It's CES 2026 Focus on Displays, Gaming Hardware, Foldables, and AI appeared first on gHacks Technology News.