Latest News - Page 412

When DeepSeek unveiled its R1 model the AI industry reeled as the company claimed it had developed an AI model that's on par with OpenAI's most-sophisticated model, but for a fraction of the cost.

But now the AI model has been out for some time, security researchers have been playing around with it and comparing it against the competition. In one set of testing, researchers from the University of Pennsylvania and hardware conglomerate Cisco pitted DeepSeek's AI against some "malicious" prompts, which are designed to bypass AI guidelines that are designed to prevent users from acquiring knowledge on how to, for example, make a bomb, generate misinformation, conduct cybercrime activities, etc.

Bypassing regulatory guidelines of a device typically called "jailbreaking," and in the instance of DeepSeek's AI, the researchers found it "failed to block a single harmful prompt." The R1 model was pitted against "50 random prompts from the HarmBench dataset," and the researchers were surprised to achieve a "100 percent attack success rate." According to the blog post, the researchers say the R1 model test results contrast starkly against other established AI models from OpenAI, Google, and Microsoft.

The FBI reports that victims lose approximately a billion dollars every year to tech support and pop-up scams. Microsoft's January 27 update to their browser Edge aims to reduce that figure.

The update in question includes a 'scareware blocker', which aims to combat a particular kind of tech support scam that involves malicious parties posing as tech support. The scam works by rushing the user with aggressive pages and using fear tactics to incite panic. These tactics include computer-generated voices, forced fullscreen effects, or hijacking of the keyboard and mouse. From there, the victim is prompted to call a fake tech-support number, after which the scam proceeds from there.

This type of scam is particularly convincing and often targets vulnerable users such as the elderly. Its prevalence is so common, in fact, that it was made into the central plot for the 2024 action film The Beekeeper. The film stars Jason Statham, who sets off on a revenge mission after his landlady becomes the victim of a multi-million dollar phishing scam.

Google continues to enhance spreadsheet capabilities with its latest Gemini upgrade for Google Sheets. The update introduces a range of AI-powered improvements, focusing on data analysis, visualization, and interpretation.

As announced on January 29th in Google's Workspace Updates blog, the features continue to build upon an existing repertoire of Gemini AI integrations with Google's apps and collaboration tools. The additions include the ability to generate charts, trends, and insights from your data and the ability to ask questions and analyze data using Python or formulas.

One key benefit of these features is the ability to derive insights and visualizations without manual data crunching. For a marketing manager, this would mean asking the tool, "What are my top-performing channels by conversion rate?" From there, Gemini would break down trends, demonstrate which platforms drive the most ROI, and visualize them with bar graphs.

With NVIDIA launching its new GeForce RTX 50 Series last week with the GeForce RTX 5080 and GeForce RTX 5090, stock shortages and unavailability meant that very few got their hands on a shiny new RTX Blackwell card for a weekend of PC gaming. With reports of hundreds of people lining up at stores and then only a handful getting GPUs, widespread reports of it being a "paper launch" have dominated the discussion.

However, as we reported last year in the lead-up to the RTX 50 Series, we've also seen cards like the GeForce RTX 4080 SUPER, GeForce RTX 4070 Ti SUPER, and even GeForce RTX 4070 series cards become increasingly more challenging to find in the US. Online retailer Newegg's RTX 4080 SUPER stock is currently being sold for over $1,500 due to stock coming from overseas.

This had led to a (temporary or not) vacuum in the mid-range and enthusiast-class market for AMD's Radeon line-up to make a move. Five of the top ten best-selling GPUs on Newegg are Radeon GPUs, including the Radeon RX 7900 XT and the Radeon RX 7800 XT - with a few highlights sold for less than their original MSRP.

The new SuperStation One is described as the "world's first affordable FPGA gaming console. " It's a faithful recreation of the original PlayStation console and hardware from the 1990s. FPGA, or Field-Programmable Gate Array hardware, is not emulation but technology that lets modders and designers recreate classic hardware with improved accuracy and the ability to run games natively.

Although the SuperStation One lacks a CD-ROM drive and doesn't ship with any copyrighted material, games, or software, users can load "backups" of PS1 games and ROMs via the microSD slot or USB drive. The SuperDock add-on supports traditional PlayStation discs. The new "console" is similar to the Analogue 3D hardware we reported on last year, another emulation-free FPGA device capable of playing Nintendo 64 games.

The SuperStation One supports modern displays with up to 1563p or 1440p output over HDMI and classic CRT TVs via composite, RCA, and component connection. It also has built-in Wi-Fi and Bluetooth for networking and controller support, Ethernet, three USB Type-A ports, and traditional PlayStation Memory Card and controller ports.

A research firm has found after conducting an extensive analysis that DeepSeek spent much more to train its R1 model than the touted $5 million.

Research firm SemiAnalysis looked at the new AI model that shook up the tech industry when it was announced, as the Chinese creators, DeepSeek, stated it only cost $6 million to create a model that was on par with OpenAI's most advanced o1 model. The announcement of the cost of DeepSeek's R1 model threw the tech space into a tailspin, wiping approximately $1 trillion away from the stock market and $600 billion from NVIDIA. For comparison, OpenAI's "GPT-4 model" is estimated to have cost anywhere between $100 - $200 million to train.

After DeepSeek's announcement, the market sentiment was the cost of training new AI models was completely overvalued, but now, according to SemiAnalysis, DeepSeek's R1 model cost the company a lot more than $5 million. The report states that DeepSeek purchased 10,000 units of NVIDIA's A100 GPUs in 2021, then acquired 10,000 "China-specific" NVIDIA H8000 AI GPUs and an additional 10,000 NVIDIA H100 GPUs. All of these GPUs, and training costs translates to $1.6 billion in capital expenditures for DeepSeek, with an estimated running cost of approximately $944 million.

Senior officials at the Office of Personnel Management (OPM) have suddenly lost access to critical department databases, which is reportedly part of Elon Musk's Department of Government Efficiency (DOGE) plan to slash government spending through downsizing.

The report from Reuters cites two anonymous federal agents who told the publication that OPM officials were locked out of a database that contains the personal data of millions of federal employees. The report states that aides to Elon Musk initiated the lockout of the officials at OPM and that the movie is part of a broader government overhaul aimed at reducing bureaucracy and implementing cost-cutting strategies.

One of the systems is called the Enterprise Human Resources Integration, and it contains dates of birth, Social Security numbers, appraisals, home addresses, pay grades, and length of service of government workers. According to the unnamed officials, OPM workers are still able to log on and access functions such as email, but they are no longer able to see the massive database that contains information on every federal worker. The unnamed officials said that since OPM employees have "no visibility" of the database, and that creating "great concern" for "real cybersecurity and hacking implications."

Intel has taken to social media to announce a new milestone achievement for its XeSS AI upscaling technology: over 150 games now include the option to turn it on. Intel Xe Super Sampling is the company's answer to NVIDIA DLSS and AMD FSR, taking an AI approach on Arc graphics hardware similar to DLSS for GeForce RTX.

As we saw in our review of the new Intel Arc B580 GPU, the first second-gen Battlemage desktop card to launch, XeSS was available in most games in our benchmark suite. Some recent titles that have added support for XeSS 2 include STALKER 2, F1 24, and Marvel Rivals. Intel XeSS 2, the latest version, also includes Frame Generation and Xe Low Latency for Arc owners.

XeSS isn't limited to desktop GPUs, as mobile Intel chips with integrated Arc graphics can also enable the technology to boost performance.

Elden Ring: Nightreign will incorporate some roguelike elements, but the director of the game has reassured fans the studio simply isn't "chasing a trend" -- there are reasons for the adjustments of in-game mechanics and gameplay.

In a recent interview with PC Gamer, Elden Ring: Nightreign game director Junya Ishizaki touched on how the latest game from FromSoftware will be different from the other titles the studio has produced. For those that don't know, Nightreign is a multiplayer-focussed title that players will be able to enjoy solo or in a group of three. The title will be "session-based," meaning players will hop in and do a "run" through enemy bosses, collect loot, and rinse and repeat.

At face value this sounds a lot like a roguelike experience, which is extremely popular with titles such as Hades and Deadcells. However, this isn't the case with Nightreign. According to Ishizaki, Nightreign has implemented some roguelike elements but with the goal of condensing down the RPG elements, the exploration, character building and leveling -- "We wanted that to feel like it was all coming together and culminating in a boss fight but in a more concentrated form."

The official artbook for Assassin's Creed Shadows has leaked, revealing character backstories, in-game weapons, locations, and more.

Over the weekend, Assassin's Creed Shadows was hit with a leak that contains spoilers for the upcoming game scheduled to release on March 20. The leak consisted of more than 200 images of the title's official artbook called "The Art of Assassin's Creed". The artbook was leaked onto a hentai forum, and the images definitely contain spoilers for in-game content as they reveal pivotal character backstories, weapons, locations, settings, and more. It should be noted the below images aren't spoiler free -- view with caution!

As for specifics, the concept art of the major cities such as Kyoto and Osaka have been revealed, along with character designs for Hattori Hanzo, Toyotomi Hideyoshi, and Momochi Sandayu. Additionally, we have looks at Portuguese Templars that are named Nuno Caro and Duarte de Melo. Furthermore, other key historical in-game figures include Kamiizumi Nobutsuna, Ashikaga Yoshiaki, Lady Nene, and Lady Oichi.

NVIDIA's cloud gaming service GeForce NOW exited Beta in February 2020, and so this month, it's celebrating its fifth anniversary as a game changer for how PC gamers access their libraries. GeForce NOW has only grown from strength to strength in recent years, with subscribers able to access their PC game library - with over 2,000 titles supported.

GeForce NOW is widely considered the best cloud gaming platform currently available. It offers up to GeForce RTX 4080 levels of performance in 4K with DLSS, ray-tracing, and even NVIDIA G-SYNC. NVIDIA notes that 17 titles will hit the service in February, with "a major game release" set to debut each week.

These major games will be the highly anticipated Kingdom Come: Deliverance II from Warhorse Studios, Avowed from Obsidian, and Sid Meier's Civilization VII from 2K. Here's the complete list of games that will be added to the GeForce NOW this month.

With the RTX Blackwell series launch last week, led by the new GeForce RTX 5080 and GeForce RTX 5090, one of the biggest stories from the event has been the lack of adequate stock to match the demand for NVIDIA's latest generation of desktop graphics cards. There have been several stories surrounding retailers getting very few cards for day one, with some claiming single-digit GeForce RTX 5090 inventory for an entire country.

This has led to many online investigations and reports calling the GeForce RTX 50 Series debut a "paper launch." And the situation, at least in the UK, is worsening. Retailer Overclockers UK has confirmed that the GeForce RTX 5080 and GeForce RTX 5090 are sold out and that, based on "high demand and limited stock," it will no longer accept pre-orders.

With that, it's currently working through a large list of pre-orders for both the RTX 5080 and RTX 5090. When it comes to NVIDIA's latest flagship, Overclockers UK notes that the current ETA for orders to be fulfilled is anywhere between 3 to 16 weeks - yeah, up to four months.

With the GeForce RTX 5070 and RTX 5070 Ti expected to launch sometime later this month, we've now got word on when we can expect to see the mainstream GeForce RTX 5060 and RTX 5060 Ti combo. Both GPUs could arrive sooner than expected, with information from Chinese hardware maker and NVIDIA partner Colorful pointing to a March 2025 release.

According to a leaked slide from @wxnod on X (via Videocardz), Taiwanese company Chaintech has outlined its roadmap for the GeForce RTX 50 Series, showing the GeForce RTX 5060 and GeForce RTX 5060 Ti launching in March. Chaintech is one of Colorful's leading suppliers of GPU components, and Colorful is one of the region's largest GPU brands.

A March 2025 launch for the GeForce RTX 5060 and RTX 5060 Ti would see NVIDIA launch its entire GeForce RTX 50 Series lineup in three months - a first. This news follows a story last week that indicated that NVIDIA could launch two GeForce RTX 5060 Ti models - one with 8GB and one with 16GB of GDDR7 memory.

The Steam Hardware and Software Survey results for January 2025 are in, and as we head into a new year, the two most popular desktop gaming GPUs are still the GeForce RTX 3060 and GeForce RTX 4060. NVIDIA's mainstream desktop cards dominate the top thirty, with no discrete options from either AMD or Intel making the list until we get to the Radeon RX 6600.

In fact, according to Valve's latest hardware survey, more PC gamers have a GeForce RTX 4090 than any single Radeon GPU. This is an incredible stat for a Titan-class or high-end enthusiast GPU that is purely a luxury item for gamers. It's also a reminder that AMD faces a monumental uphill battle to reclaim market share and mindshare among gamers.

In a sea of green, AMD will launch its upcoming RDNA 4 line-up of mid-range offerings in March. With its Radeon RX 9070 and RTX 9070 XT combo, AMD will target the mid-range market dominated by the GeForce RTX 4070 and RTX 4070 SUPER. They could be successful with the correct pricing and performance; however, it will probably take a couple of generations to regain a foothold in the discrete GPU space.

NVIDIA's new GeForce RTX 5090 and GeForce RTX 5080 graphics cards are pretty much sold out US-wide, with the biggest retailers in the US sold out: Micro Center, Best Buy, and Newegg are all wiped of their stock.

No matter the price, no matter the whinging of how the custom RTX 5090 and RTX 5080 cards are 40% (or more) expensive, NVIDIA's new graphics cards have been wiped from shelves. There were rumors before the launch of how bad the supply situation of the new RTX 50 series GPU was, but it seems to be even worse than that.

US retailers including Micro Center, Newegg, and Best Buy might be completely out of stock, but if PC gamers still wanted to find themselves a new GeForce RTX 5090 or RTX 5080 graphics card... there are still options, but it'll cost ya. You'd need to buy a pre-built gaming PC with a new RTX 5090 or RTX 5080 inside. It'll be more expensive, but you'll have a new Blackwell GPU.

NVIDIA's next-gen GeForce RTX 60 series "Rubin" GPUs are now officially in the headlines, with leakers creating a wishlist of things they'd like to see from the RTX 60 series.

In a new post on X by leaker "harukaze5719" and followed up by 3DCenter.org, we've got a mighty nice list of things to expect from NVIDIA's next-gen GeForce RTX 60 series GPUs. NVIDIA's new GeForce RTX 60 series should use the new Rubin GPU architecture, with a 10%+ performance lift in rasterization and 20%+ performance lift in RT on the wishlist. Not only that, but next-gen DLSS 5 upscaling with a 2x performance uplift compared to DLSS 4.

In a reply to that post, 3DCenter added that we should see the next-gen GeForce RTX 60 series fabbed on TSMC's new 3nm process node, with a bigger 50%+ lift in RT performance, "maybe just a small/mediocre raster jump" and more VRAM on all RTX 60 series GPUs. 3DCenter wants to see more AI generation on parts of the frames, noting that "usually after a weaker gen a stronger gen will follow".

Samsung Electronics has posted its Q4 2024 earnings, pulling in $54.18 billion in revenue, with an operating profit of $4.65 billion, with some new details going on with its troubled HBM memory.

Samsung says that it started supplying 8-Hi and 12-Hi HBM3E memory in Q3 2024, and that revenue from HBM3E has surpassed HBM3. The South Korean giant is still working on an "optimized version" of HBM3E that will ship in Q1 2025, while NVIDIA CEO Jensen Huang has said Samsung needed to redesign its HBM3E memory and the "optimized version" is probably just that, redesigned HBM3E.

The company says that HBM3E might see temporary demand and price drops in Q1 2025, as AI clients wait for the new "optimized version" instead of taking delivery of the older one. Insider Dan Nystedt notes that "no other major memory chip maker uses garbage terms like 'optimized version' for HBM products'

AMD's new flagship Ryzen 9 9950X3D and Ryzen 9 9900X3D processors are rumored to drop by the end of March 2025, after previous rumors had the new X3D chips pegged for a late January launch... as that time has come and passed, and the new X3D chips aren't here.

In a new report from French outlet Cowcotland, the site reports from their sources that AMD's new Ryzen 9 9950X3D and 9900X3D processors will be launching by the end of March. Not only that, but it seems AMD will be lining up the launch of its new X3D processors with its new RDNA 4-powered Radeon RX 9070 XT and Radeon RX 9070 graphics cards.

Cowcotland reported: "Regarding pricing, we have no information. However, our sources indicate that AMD's two new X3D models are expected to arrive by the end of March, around the same time as the new Radeon RX 9070 and RX 9070 XT".

GALAX has launched its flagship GeForce RTX 5080 HOF OC Lab Plus-X Edition graphics card, with a gorgeous white PCB, but not in RTX 5090 form which is disappointing to say the least. Anyway, let's check it out:

The folks over at VideoCardz have the scoop, showing the world the new GALAX GeForce RTX 5080 HOF OC Lab Plus-X graphics card with its 12-layer PCB and 14+6+6 phase power design. GALAX powers the card with 70A high-current DR-MOS and powered by a single 16-pin power connector.

We've got a triple-fan cooling design with two of the fans at 102mm and one of them at 92mm, with GALAX using "Frost Ring Blades" that have been designed to increase air pressure and reduce noise. GALAX has factory overclocking of its new GeForce RTX 5080 HOF OC Lab Plus-X Edition graphics card, with an overclock of up to 2730MHz (I'm sure it can be manually pushed much further than this) and a dual BIOS setting that enables this power setting.

ASUS has shared pricing details for the full range of its RTX 5090 and 5080 graphics cards, and the top-end model of the Blackwell flagship exceeds the $3,000 mark in a move that has definitely caused a stir.

This is the ASUS ROG Astral LC GeForce RTX 5090 OC Edition - a clear argument for the product name length having a direct correlation to more expensive pricing - and it's official asking price is $3,100.

NVIDIA's MSRP for the RTX 5090 is $2,000, so with this card, ASUS is blowing past that recommendation by 55%. And yes, of course we expect to pay a significant amount more for premium third-party graphics cards, but the ROG Astral LC version appears to be taking the cake (or indeed the cake, its stand, and the table beneath).