La période des fêtes est arrivée, et GeForce NOW clôture l'année 2025 avec une gamme de jeux en cloud très riche, comprenant 30 nouveaux titres, dont Hogwarts Legacy et la collection LEGO Harry Potter.
GeForce NOW ajoute également les premiers titres Activision via Ubisoft+ Premium, notamment les jeux d'action Call of Duty: Modern Warfare II et III, Crash Bandicoot N. Sane Trilogy et Spyro Reignited Trilogy, faisant de GeForce NOW la première plateforme de cloud gaming à proposer ces titres Ubisoft+ Premium Activision, tous diffusés en streaming depuis des serveurs cloud équipés de GeForce RTX.
Que la neige tombe, mais ne laissez pas passer cette offre : la promotion Half Price Holiday est en ligne cette semaine et offre 50 % de réduction sur le premier mois d'abonnement Performance ou Ultimate. C'est donc le moment idéal pour transformer n'importe quel appareil en une plateforme de jeu haut de gamme afin de poursuivre votre série de jeux hivernaux.
Mais ce n'est pas tout : grâce à l'authentification unique Battle.net, il est désormais plus facile que jamais de se lancer dans Overwatch 2, Diablo IV et d'autres titres pris en charge avec une seule connexion sur tous les appareils, tandis que le concours vidéo communautaire GeForce NOW invite les joueurs à partager leurs meilleurs moments des fêtes pour tenter de gagner un abonnement Ultimate d'un an, jusqu'au 7 décembre. […]
An anonymous reader quotes a report from Ars Technica: A Donald Trump-backed push has failed to wedge a federal measure that would block states from passing AI laws for a decade into the National Defense Authorization Act (NDAA). House Majority Leader Steve Scalise (R-La.) told reporters Tuesday that a sect of Republicans is now "looking at other places" to potentially pass the measure. Other Republicans opposed including the AI preemption in the defense bill, The Hill reported, joining critics who see value in allowing states to quickly regulate AI risks as they arise.
For months, Trump has pressured the Republican-led Congress to block state AI laws that the president claims could bog down innovation as AI firms waste time and resources complying with a patchwork of state laws. But Republicans have continually failed to unite behind Trump's command, first voting against including a similar measure in the "Big Beautiful" budget bill and then this week failing to negotiate a solution to pass the NDAA measure. [...]
"We MUST have one Federal Standard instead of a patchwork of 50 State Regulatory Regimes," Trump wrote on Truth Social last month. "If we don't, then China will easily catch us in the AI race. Put it in the NDAA, or pass a separate Bill, and nobody will ever be able to compete with America." If Congress bombs the assignment to find another way to pass the measure, Trump will likely release an executive order to enforce the policy. Republicans in Congress had dissuaded Trump from releasing a draft of that order, requesting time to find legislation where they believed an AI moratorium could pass. "The controversial proposal had faced backlash from a nationwide, bipartisan coalition of state lawmakers, parents, faith leaders, unions, whistleblowers, and other public advocates," the NDAA, a bipartisan group that lobbies for AI safety laws, said in a press release.
This "widespread and powerful" movement "clapped back" at Republicans' latest "rushed attempt to sneak preemption through Congress," Brad Carson, ARI's president, said, because "Americans want safeguards that protect kids, workers, and families, not a rules-free zone for Big Tech."
alternative_right quotes a report from the Guardian: The statue looms and glints at more than 11 feet tall and weighing 3,500 pounds, looking out at the city with, how to put it ... a characteristically stern expression? Despite its daunting appearance and history as a crimefighter of last resort, the giant new bronze figure of the movie character RoboCop is being seen as a symbol of hope, drawing fans and eliciting selfie mania since it began standing guard over Detroit on Wednesday afternoon. It has been 15 years in the making. Even in a snowstorm in the dark, people were driving by to see it, said Jim Toscano, co-owner of the Free Age film production company, where the statue now stands firmly bolted down near the sidewalk. RoboCop hit theaters in 1987, portraying a near-future Detroit as crime-ridden and poorly protected by a beleaguered and outgunned police force, until actor Peter Weller appeared as a nearly invincible cyborg, apparently created by a nefarious corporation bent on privatizing policing. A grassroots campaign to build a RoboCop statue in Detroit began in 2010, eventually raising over $67,000 on Kickstarter and resulting in a completed sculpture in 2017. However, hosting setbacks caused it to get stuck, "stored away from public view," reports the Guardian. The project finally found a home after business owner Mike Toscano agreed to display it in their new open-air product market, calling it "too unique and too cool not to do."
The big set of kernel graphics driver features were merged today for the Linux 6.19 kernel. As usual there is a lot of new feature work on the AMD Radeon, Intel, and NVIDIA graphics drivers plus the smaller Arm/embedded graphics like now having initial Qualcomm Gen8 GPU support. Plus the growing number of accelerator "accel" drivers for NPUs / AI accelerators...
U.S. regulators are pressing Waymo for answers after Texas officials reported 19 instances of its self-driving cars illegally passing stopped school buses, including cases that occurred after Waymo claimed to have deployed a software fix. Longtime Slashdot reader BrendaEM shares the report from Reuters: In a November 20 letter posted by NHTSA, the Austin Independent School District said five incidents occurred in November after Waymo said it had made software updates to resolve the issue and asked the company to halt operations around schools during pick-up and drop-off times until it could ensure the vehicles would not violate the law. "We cannot allow Waymo to continue endangering our students while it attempts to implement a fix," a lawyer for the school district wrote, citing one incident involving a Waymo that was "recorded driving past a stopped school bus only moments after a student crossed in front of the vehicle, and while the student was still in the road."
The letter prompted NHTSA to ask Waymo on November 24 if it would comply with the request to cease self-driving operations during student pick-up and drop-off times, adding: "Was an appropriate software fix implemented or developed to mitigate this concern? And if so, does Waymo plan to file a recall for the fix?" The school district told Reuters on Thursday that Waymo refuses to halt operations around schools and said another incident involving a self-driving car and an actively loading school bus occurred on December 1, which "indicates that those programming changes did not resolve the issue or our concerns."
In a statement, Waymo did not answer why it had refused to halt operations around Austin schools or answer if it would issue a recall. "We're deeply invested in safe interaction with school buses. We swiftly implemented software updates to address this and will continue to rapidly improve," Waymo said. NHTSA said in a letter to Waymo on Wednesday that it was demanding answers to a series of questions by January 20 about incidents involving school buses and details of software updates to address safety concerns.
Ancient Slashdot user Alain Williams shares a report from Al Jazeera: The Irish Council for Civil Liberties (ICCL) has announced it filed a complaint against Microsoft, accusing the global tech giant of unlawfully processing data on behalf of the Israeli military and facilitating the killings of Palestinian civilians in Gaza. In the complaint, the council asked the Data Protection Commission -- the European Union's lead data regulator for the company -- to "urgently investigate" Microsoft Ireland's processing.
"Microsoft's technology has put millions of Palestinians in danger. These are not abstract data-protection failures -- they are violations that have enabled real-world violence," Joe O'Brien, ICCL's executive director, said in a statement. "When EU infrastructure is used to enable surveillance and targeting, the Irish Data Protection Commission must step in -- and it must use its full powers to hold Microsoft to account."
After months of complaints from rights groups and Microsoft whistleblowers, the company said in September it cancelled some services to the Israeli military over concerns that it was violating Microsoft's terms of service by using cloud computing software to spy on millions of Palestinians.
A light aircraft crashed in Gloucestershire after a 3D-printed plastic air-induction elbow softened from engine heat and collapsed, cutting power during final approach and causing the plane to undershoot the runway. Investigators say the part was made from "inappropriate material" and safety actions will be taken in the future regarding 3D printed parts. The BBC reports: Following an "uneventful local flight", the AAIB report said the pilot advanced the throttle on the final approach to the runway, and realized the engine had suffered a complete loss of power. "He managed to fly over a road and a line of bushes on the airfield boundary, but landed short and struck the instrument landing system before coming to rest at the side of the structure," the report read.
It was revealed the part had been installed during a modification to the fuel system and collapsed due to its 3D-printed plastic material softening when exposed to heat from the engine. The Light Aircraft Association (LAA) said it now intends to take safety actions in response to the accident, including a "LAA Alert" regarding the use of 3D-printed parts that will be sent to inspectors.
Amazon AWS today announced Graviton5 as their newest-generation ARM64 server processor for their EC2 cloud. Graviton5 is being promoted as offering 25% higher performance over existing Graviton4 processors...
Russia has blocked Apple's FaceTime and the gaming platform Roblox as part of a broader crackdown on foreign tech platforms. CBC News reports: Both restrictions are part of an accelerating clampdown on foreign tech platforms: In the case of FaceTime, Russian authorities allege it is being used for criminal activity, while Roblox was accused of distributing extremist materials and "LGBT propaganda." The move follows restrictions against Google's YouTube, Meta's WhatsApp and the Telegram messaging service.
Critics say the curbs amount to censorship and a tightening of state control over private communications. Russia says they are legitimate law enforcement measures. Russian authorities have this year launched a state-backed rival app called Max, which critics say could be used for surveillance -- allegations that state media have dismissed as false.
Justifying its decision, the communications regulator, Roskomnadzor, said in an emailed statement: "According to law enforcement agencies, FaceTime is being used to organize and carry out terrorist attacks in the country, recruit perpetrators, and commit fraud and other crimes against Russian citizens." The watchdog did not cite evidence in support of the allegations.
NVIDIA just released CUDA 13.1 for what they claim is "the largest and most comprehensive update to the CUDA platform since it was invented two decades ago." The most notable addition with the CUDA 13.1 release is CUDA Tile as a new tile-based programming model...
The EU has opened an antitrust investigation into Meta over a new WhatsApp policy that could block rival AI assistants from accessing the platform. Complaints from smaller AI developers triggered the probe, which could lead to fines of up to 10% of Meta's global revenue if the company is found to have abused its dominance. Reuters reports: EU antitrust chief Teresa Ribera said the move was to prevent dominant firms from "abusing their power to crowd out innovative competitors." She added interim measures could be imposed to block Meta's new WhatsApp AI policy rollout. "AI markets are booming in Europe and beyond," she said. "This is why we are investigating if Meta's new policy might be illegal under competition rules, and whether we should act quickly to prevent any possible irreparable harm to competition in the AI space."
A WhatsApp spokesperson called the claims "baseless," adding that the emergence of chatbots on its platforms had put a "strain on our systems that they were not designed to support," a reference to AI systems from other providers. "Still, the AI space is highly competitive and people have access to the services of their choice in any number of ways, including app stores, search engines, email services, partnership integrations, and operating systems."
The many hardware monitoring (HWMON) subsystem updates were merged today for Linux 6.19 that is predominantly around delivering new hardware support...
Longtime Slashdot reader williamyf writes: The Italian company Bending Spoons seems to be on an acquisitions spree. Their recent acquisitions of AOL and Vimeo are not yet finalized, yet on Dec. 2 they announced they are buying Eventbrite, a company specializing in publicizing and organizing local events, for just half a milliard USD. Bending Spoons' portfolio also includes other companies like Evernote and WeTransfer. Further reading: Private Equity Hipsters Are Coming For Your Favorite Apps (2024)
joshuark shares a report from BleepingComputer: Microsoft has silently "mitigated" a high-severity Windows LNK vulnerability exploited by multiple state-backed and cybercrime hacking groups in zero-day attacks. Tracked as CVE-2025-9491, this security flaw allows attackers to hide malicious commands within Windows LNK files, which can be used to deploy malware and gain persistence on compromised devices. However, the attacks require user interaction to succeed, as they involve tricking potential victims into opening malicious Windows Shell Link (.lnk) files. Thus some element of social engineering, and user technically naive and gullibility such as thinking Windows is secure is required. [...]
As Trend Micro threat analysts discovered in March 2025, the CVE-2025-9491 was already being widely exploited by 11 state-sponsored groups and cybercrime gangs, including Evil Corp, Bitter, APT37, APT43 (also known as Kimsuky), Mustang Panda, SideWinder, RedHotel, Konni, and others. Microsoft told BleepingComputer in March that it would "consider addressing" this zero-day flaw, even though it didn't "meet the bar for immediate servicing." ACROS Security CEO and 0patch co-founder Mitja Kolsek found, Microsoft has silently changed LNK files in the November updates in an apparent effort to mitigate the CVE-2025-9491 flaw. After installing last month's updates, users can now see all characters in the Target field when opening the Properties of LNK files, not just the first 260. As the movie the Ninth Gate stated: "silentium est aurum"
Comme nous le rappelions pas plus tard qu'hier avec la mésaventure qu'est en train de connaitre la marque Transcend, la NAND est elle aussi en train d'être touchée par la pénurie occasionnée par la demande gigantesque du secteur de l'IA. Ceux qui s'en frottent les mains, ce sont bien sûr les fabrica...
A severe spike in global DRAM prices has pushed Samsung Semiconductor to refuse a long-term RAM order from its own sibling, Samsung Electronics. The move is forcing the smartphone division into short, expensive renegotiations, which will likely mean higher costs for consumer devices. PCWorld reports: Samsung subsidiaries are, naturally, going to look to Samsung Semiconductor first when they need parts. Such was reportedly the case for Samsung Electronics, in search of memory supplies for its newest smartphones as the company ramps up production for 2026 flagship designs. But with so much RAM hardware going into new "AI" data centers -- and those companies willing to pay top dollar for their hardware -- memory manufacturers like Samsung, SK Hynix, and Micron are prioritizing data center suppliers to maximize profits.
The end result, according to a report from SE Daily spotted by SamMobile, is that Samsung Semiconductor rejected the original order for smartphone DRAM chips from Samsung Electronics' Mobile Experience division. The smartphone manufacturing arm of the company had hoped to nail down pricing and supply for another year. But reports say that due to "chipflation," the phone-making division must renegotiate quarterly, with a long-term supply deal rejected by its corporate sibling. A short-term deal, with higher prices, was reportedly hammered out.
The out-of-tree Bcachefs file-system is ready with its reconcile feature, which previously was known as "rebalance_v2", and what lead developer Kent Overstreet calls the biggest feature to this copy-on-write file-system in the last two years...
An anonymous reader quotes a report from the Guardian: Almost three in 10 GPs in the UK are using AI tools such as ChatGPT in consultations with patients, even though it could lead to them making mistakes and being sued, a study reveals. The rapid adoption of AI to ease workloads is happening alongside a "wild west" lack of regulation of the technology, which is leaving GPs unaware which tools are safe to use. That is the conclusion of research by the Nuffield Trust thinktank, based on a survey of 2,108 family doctors by the Royal College of GPs about AI and on focus groups of GPs.
Ministers hope that AI can help reduce the delays patients face in seeing a GP. The study found that more and more GPs were using AI to produce summaries of appointments with patients, assisting their diagnosis of the patient's condition and routine administrative tasks. In all, 598 (28%) of the 2,108 survey respondents said they were already using AI. More male (33%) than female (25%) GPs have used it and far more use it in well-off than in poorer areas.
It is moving quickly into more widespread use. However, large majorities of GPs, whether they use it or not, worry that practices that adopt it could face "professional liability and medico-legal issues," and "risks of clinical errors" and problems of "patient privacy and data security" as a result, the Nuffield Trust's report says. [...] In a blow to ministerial hopes, the survey also found that GPs use the time it saves them to recover from the stresses of their busy days rather than to see more patients. "While policymakers hope that this saved time will be used to offer more appointments, GPs reported using it primarily for self-care and rest, including reducing overtime working hours to prevent burnout," the report adds.
This week brought the official release of FreeBSD 15.0 as the latest major update to this BSD operating system. In being eager to test out this new FreeBSD release, for this first round of FreeBSD 15.0 benchmarking is seeing how it compares to the former FreeBSD 14.3 release on a Supermicro + AMD EPYC Turin server.
Les grandes manœuvres financières se poursuivent sur le terrain de l’IA générative, avec une révélation de taille : Anthropic, éditeur de Claude, étudierait les modalités d’une introduction en bourse dès 2026. OpenAI aurait pour sa part décidé de retarder son offensive dans la publicité en ligne, en raison d’une consigne d’urgence passée par Sam Altman. Celui-ci s’inquiéterait en effet de l’accueil réservé à Gemini 3, que Google diffuse le plus largement possible…
Alors que la rumeur prête à OpenAI l’intention d’aller chercher le carburant nécessaire à sa frénésie d’investissements en bourse, l’éditeur de ChatGPT pourrait-il se faire damer le pion par son concurrent Anthropic ? C’est l’une des questions qui a animé les milieux financiers de la tech cette semaine, après que le Financial Times a révélé mercredi que la maison mère des modèles de langage Claude travaillait à la question avec son conseil, le cabinet Wilson Sonsini.
Anthropic sur le sentier de la bourse ?
À ce stade, il ne s’agit encore que de bruits de couloir, mais ces derniers résonnent assez fort pour que le quotidien britannique se risque à avancer cette éventualité d’une introduction en bourse, pour laquelle Dario Amodei et ses équipes auraient déjà sondé plusieurs grandes banques. Une autre source évoque un processus engagé dès 2026.
Du côté d’Anthropic, on répond qu’il est naturel, à ce niveau d’activités, d’étudier les différentes options stratégiques possibles, sans confirmer quoi que ce soit. « Nous n’avons pris aucune décision quant à une date ou même une éventuelle ouverture [du capital, ndlr] au public, et nous n’avons aucune information à partager pour le moment », indique un porte-parole.
L’entrée en bourse d’Anthropic constituerait à n’en pas douter une opération d’un montant record, d’où l’attention particulière accordée au sujet. Le dernier tour de financement réalisé par l’entreprise, mi-novembre, portait sur 15 milliards de dollars, amenés par NVIDIA et Microsoft, en échange de promesses d’achat de composants et de services cloud, conformément à la logique d’engagements circulaires qui semble désormais prévaloir dans le secteur.
Cet accord tripartite a été conclu, estime le FT, sur la base d’une capitalisation comprise entre 300 et 350 milliards de dollars pour Anthropic. Une introduction en bourse signifierait une nouvelle révision à la hausse de cette valorisation, sur un marché dont la dimension spéculative et les montages financiers hasardeux font redouter à certains l’éclatement à venir d’une bulle de l’IA…
L’entrée, à Wall Street, des deux pure players les plus emblématiques du secteur déplacerait par ailleurs leur rivalité vers un nouveau terrain de jeu, purement financier cette fois. De quoi, sans doute, soutenir un phénomène à venir de concentration dans le secteur.
Anthropic vient à ce sujet d’annoncer l’acquisition de Bun, éditeur d’un framework JavaScript open source qui sous-tend déjà le fonctionnement de certains composants de Claude Code. Reste à savoir pourquoi Anthropic n’a pas fait créer un fork de Bun à son IA, si cette dernière est si performante, s’interroge un internaute taquin.
« Code rouge » autour de ChatGPT chez OpenAI
En attendant, la bataille se joue à l’échelle commerciale bien sûr – comme en témoignent les nombreux accords annoncés entre acteurs de l’IA et poids lourds des pans plus traditionnels de l’économie –, mais aussi technique. Et sur ce point, c’est Google qui semble avoir pris l’avantage, au moins temporairement, avec la sortie de Gemini 3 Pro, dont les performances sont globalement saluées.
En matière de communication, Google s’est d’ailleurs trouvé un allié de poids, en la personne de Marc Benioff, CEO de Salesforce, qui a littéralement encensé Gemini 3 le 23 novembre dernier : « Bordel de merde. J’utilise ChatGPT tous les jours depuis 3 ans. Je viens de passer deux heures sur Gemini 3. Je ne reviendrai pas en arrière ».
Marc Benioff, VRP pas super élégant mais très efficace de Gemini 3 – capture d’écran Next
Le succès de Gemini 3 n’a pas laissé Sam Altman indifférent, c’est peu de le dire. Dans un mémo interne, envoyé lundi 1ᵉʳ décembre aux équipes d’OpenAI et révélé par The Information, le patron d’OpenAI a annoncé un « code rouge », face à ce qu’il considère comme une « situation critique » pour ChatGPT. L’alerte se traduirait par un focus accru sur le développement produit de ChatGPT, quitte à retarder ou abandonner certains projets connexes.
« Notre priorité est désormais de continuer à améliorer ChatGPT, de poursuivre sa croissance et d’étendre son accessibilité à travers le monde, tout en le rendant encore plus intuitif et personnalisé », a souligné publiquement Nick Tuley, patron de ChatGPT chez OpenAI, tout en se réjouissant des résultats accomplis au cours de ces trois premières années de mise à disposition au public.
De la pub dans ChatGPT… mais quand ?
Le code rouge prononcé par Sam Altman pourrait avoir un effet sur un des sujets les plus stratégiques qui soient pour OpenAI : la monétisation via la publicité. D’après The Information, l’intégration de publicités au sein de ChatGPT ferait en effet partie des dossiers sortis du haut de la pile, alors même que l’entreprise a besoin de justifier de revenus en hausse pour soutenir sa recherche de financements.
OpenAI n’a pas communiqué publiquement sur ses velléités en matière de publicités, mais une confirmation de ses travaux sur le sujet a fait surface fin novembre. Des internautes ont effet découvert, au sein d’une bêta de l’application Android de ChatGPT, des mentions à l’intitulé explicitement rattaché à l’affichage de réclames, telles que ads feature, bazaar content, search ad, ou search ads carousel.
Le sujet n’est pas trivial : pour être efficace, la réclame au sein de ChatGPT devra en effet être contextualisée soit en fonction des requêtes de l’utilisateur (ce qui pose des questions de neutralité dans la génération de la réponse), soit en fonction de la connaissance qu’a le modèle de l’utilisateur (et donc d’éléments relevant de sa vie privée).
Google ouvre les vannes sur Gemini 3, puis les referme (un peu)
S’il y en a dont les projets en la matière seront scrutés de près, c’est bien sûr Google, qui dispose avec son moteur de recherche, Android et Chrome, de multiples points de contact pour toucher l’internaute et lui dispenser des services d’IA générative financés par la publicité.
Et Google semble bien décidée à accélérer la conquête de nouveaux utilisateurs. Si OpenAI caracole en tête en revendiquant 800 millions d’utilisateurs par semaine sur ChatGPT, Google est de son côté de 450 millions à 650 millions d’utilisateurs actifs mensuels, selon les déclarations d’Alphabet lors de la présentation de ses derniers résultats financiers.
Pour continuer sur cette lancée, Google joue la carte de la synergie. Les abonnés à l’offre de stockage cloud Google One Premium ont ainsi reçu cette semaine une invitation à profiter sans surcoût des avantages de l’abonnement AI Plus, qui intègre un accès plus généreux à Gemini 3 Pro et Nano Banana pour les images. La dernière version du modèle s’est également invitée dans l’AI Mode, l’agent conversationnel de recherche de Google (encore non disponible en France) qui constitue une alternative à la traditionnelle liste de liens.
Problème : gagner des utilisateurs coûte cher, et tous ne basculent pas vers des formules payantes, ce qui conduit dans le même temps Google, comme OpenAI, à limiter en volume les possibilités offertes en version gratuite. Chez OpenAI, c’est au niveau de Sora, le modèle de génération vidéo, que la barre est abaissée à six requêtes par jour, contre trente précédemment. « Nos cartes graphiques surchauffent, et nous voulons permettre au plus grand nombre de personnes possible d’accéder à Sora ! », explique Bill Peebles, patron de Sora, dans un message du 28 novembre.
La veille, Google avait procédé à un mouvement similaire, en réduisant les paliers d’utilisation gratuite de Gemini 3 Pro et de Nano Banana.
Connexion
