The weight of AI server racks has reached a point where legacy data centers cannot accommodate them even with significant retrofitting efforts, The Verge reports. Chris Brown, chief technical officer at Uptime Institute, said most retrofitting attempts would require "bulldozing the building and starting over from scratch."

AI racks are projected to reach 5,000 pounds compared to the 400 to 600 pounds that racks weighed three decades ago. The dramatic increase stems from hundreds to 1,000 GPUs packed densely into each rack alongside memory chips and liquid cooling systems that can add substantial weight. AI workloads now consume up to 350 kilowatts per rack, 35 times the 10 kilowatts that traditional computer chip workloads averaged a decade ago. Legacy data centers with raised floors typically max out at around 1,250 pounds per square foot for static loads.

Chris McLean, president of Critical Facility Group, said that rack heights have grown from 6 feet to 9 feet over nearly two decades, creating problems with doorframes and freight elevators in older buildings.

A millisecond used to be a big deal for the world's quickest traders. A dispute over huge trading profits at one of the world's largest futures exchanges shows they now think a million times faster [non-paywalled source]. From a report: The controversy is about an arcane technical maneuver in which high-speed traders bombard Frankfurt-based Eurex with useless data. The idea is to keep their connections to the exchange warm so they can react fractionally faster to market-moving information. The battle is the latest chapter in a decadeslong contest among secretive ultrafast trading firms, which have pursued a relentless quest for minuscule speed advantages.

A group of high-frequency trading firms has exploited the practice to rake in hundreds of millions of dollars, says Mosaic Finance, a French firm that has complained to Eurex and European regulators. "An arms race is OK, but you must use legal weapons," said Hugues Morin, founder of Mosaic. Eurex says Mosaic's claims are baseless.

[...] High-speed traders often seek to capture fleeting differences between prices of related assets, making quick response times critical. If benchmark Euro Stoxx 50 index futures rise, for example, contracts tied to Germany's DAX will usually follow. A first mover will be able to buy DAX futures before they tick higher, then sell out at a higher price -- a strategy that can add up to big profits over time. The maneuver that prompted Mosaic's spat with Eurex can improve reaction times by about 3.2 nanoseconds, according to the French firm, which calls it "corrupted speculative triggering," or CST for short.

An anonymous reader quotes a report from BleepingComputer: Audio streaming platform SoundCloud has confirmed that outages and VPN connection issues over the past few days were caused by a security breach in which threat actors stole a database containing user information. The disclosure follows widespread reports over the past four days from users who were unable to access SoundCloud when connecting via VPN, with attempts resulting in the site displaying 403 "forbidden" errors.

In a statement shared with BleepingComputer, SoundCloud said it recently detected unauthorized activity involving an ancillary service dashboard and activated its incident response procedures. SoundCloud acknowledged that a threat actor accessed some of its data but said the exposure was limited in scope. [...] BleepingComputer has learned that the breach affects 20% of SoundCloud's users, which, based on publicly reported user figures, could impact roughly 28 million accounts. The company said it is confident that all unauthorized access to SoundCloud systems has been blocked and that there is no ongoing risk to the platform. "We understand that a purported threat actor group accessed certain limited data that we hold," SoundCloud told BleepingComputer. "We have completed an investigation into the data that was impacted, and no sensitive data (such as financial or password data) has been accessed. The data involved consisted only of email addresses and information already visible on public SoundCloud profiles."

An anonymous reader quotes a report from Ars Technica: Microsoft is killing off an obsolete and vulnerable encryption cipher that Windows has supported by default for 26 years following more than a decade of devastating hacks that exploited it and recently faced blistering criticism from a prominent US senator. When the software maker rolled out Active Directory in 2000, it made RC4 a sole means of securing the Windows component, which administrators use to configure and provision fellow administrator and user accounts inside large organizations. RC4, short for Rivist Cipher 4, is a nod to mathematician and cryptographer Ron Rivest of RSA Security, who developed the stream cipher in 1987. Within days of the trade-secret-protected algorithm being leaked in 1994, a researcher demonstrated a cryptographic attack that significantly weakened the security it had been believed to provide. Despite the known susceptibility, RC4 remained a staple in encryption protocols, including SSL and its successor TLS, until about a decade ago. [...]

Last week, Microsoft said it was finally deprecating RC4 and cited its susceptibility to Kerberoasting, the form of attack, known since 2014, that was the root cause of the initial intrusion into Ascension's network. "By mid-2026, we will be updating domain controller defaults for the Kerberos Key Distribution Center (KDC) on Windows Server 2008 and later to only allow AES-SHA1 encryption," Matthew Palko, a Microsoft principal program manager, wrote. "RC4 will be disabled by default and only used if a domain administrator explicitly configures an account or the KDC to use it." [...] Following next year's change, RC4 authentication will no longer function unless administrators perform the extra work to allow it. In the meantime, Palko said, it's crucial that admins identify any systems inside their networks that rely on the cipher. Despite the known vulnerabilities, RC4 remains the sole means of some third-party legacy systems for authenticating to Windows networks. These systems can often go overlooked in networks even though they are required for crucial functions.

To streamline the identification of such systems, Microsoft is making several tools available. One is an update to KDC logs that will track both requests and responses that systems make using RC4 when performing requests through Kerberos. Kerberos is an industry-wide authentication protocol for verifying the identities of users and services over a non-secure network. It's the sole means for mutual authentication to Active Directory, which hackers attacking Windows networks widely consider a Holy Grail because of the control they gain once it has been compromised. Microsoft is also introducing new PowerShell scripts to sift through security event logs to more easily pinpoint problematic RC4 usage. Microsoft said it has steadily worked over the past decade to deprecate RC4, but that the task wasn't easy. "The problem though is that it's hard to kill off a cryptographic algorithm that is present in every OS that's shipped for the last 25 years and was the default algorithm for so long, Steve Syfuhs, who runs Microsoft's Windows Authentication team, wrote on Bluesky. "See," he continued, "the problem is not that the algorithm exists. The problem is how the algorithm is chosen, and the rules governing that spanned 20 years of code changes."

A critical React vulnerability (CVE-2025-55182) is being actively exploited at scale by Chinese, Iranian, North Korean, and criminal groups to gain remote code execution, deploy backdoors, and mine crypto. The Register reports: React maintainers disclosed the critical bug on December 3, and exploitation began almost immediately. According to Amazon's threat intel team, Chinese government crews, including Earth Lamia and Jackpot Panda, started battering the security hole within hours of its disclosure. Palo Alto Networks' Unit 42 responders have put the victim count at more than 50 organizations across multiple sectors, with attackers from North Korea also abusing the flaw.

Google, in a late Friday report, said at least five other suspected PRC spy groups also exploited React2Shell, along with criminals who deployed XMRig for illicit cryptocurrency mining, and "Iran-nexus actors," although the report doesn't provide any additional details about who the Iran-linked groups are and what they are doing after exploitation. "GTIG has also observed numerous discussions regarding CVE-2025-55182 in underground forums, including threads in which threat actors have shared links to scanning tools, proof-of-concept (PoC) code, and their experiences using these tools," the researchers wrote.

A Kansas man who sued Verizon in small claims court after the carrier refused to unlock his iPhone has won his case, scoring a small but meaningful victory against a company that retroactively applied a policy change to deny his unlock request.

Patrick Roach bought a discounted iPhone 16e from Verizon's Straight Talk brand in February 2025, intending to pay for one month of service before switching the device to US Mobile. Under FCC rules dating back to a 2019 waiver, Verizon must unlock phones 60 days after activation on its network. Verizon refused to unlock the phone, citing a new policy implemented on April 1, 2025 requiring "60 days of paid active service."

Roach had purchased his device over a month before that policy took effect. Magistrate Judge Elizabeth Henry ruled in October 2025 that applying the changed terms to Roach's earlier purchase violated the Kansas Consumer Protection Act. The court ordered Verizon to refund Roach's $410.40 purchase price plus court costs. Roach had previously rejected a $600 settlement offer because it would have required him to sign a non-disclosure agreement. He estimated spending about 20 hours on the lawsuit but said "it wasn't about" the money.

Google has decided to retire its free dark web monitoring tool, saying it wasn't as helpful as the company hoped. From a report: In a support page, Google announced the discontinuation of the "dark web report" tool, two years after offering it as a free perk to Gmail users before expanding it more broadly. The feature worked by scanning for your email addresses to determine whether they had appeared in data breaches, which often circulate on Dark Web marketplaces. The tool could then alert you about where the data was exposed, including any accompanying details such as dates of birth, addresses, and phone numbers.

LG smart TV owners discovered over the weekend that a recent webOS software update had quietly installed Microsoft Copilot on their devices, and the app cannot be uninstalled. Affected users report the feature appears automatically after installing the latest webOS update on certain models, sitting alongside streaming apps like Netflix and YouTube.

LG's support documentation confirms that certain preinstalled or system apps can only be hidden, not deleted. At CES 2025, LG announced plans to integrate Copilot into webOS as part of its "AI TV" strategy, describing it as an extension of its AI Search experience. The current implementation appears to function as a shortcut to a web-based Copilot interface rather than a native application. Samsung TVs include Google's Gemini in a similar fashion. Users wanting to avoid the feature entirely are left with one option: disconnecting their TV from the internet.

The Black Hat Europe hacker conference in London included a session titled "Don't Judge an Audiobook by Its Cover" about a two critical (and now fixed) flaws in Amazon's Kindle. The Times reports both flaws were discovered by engineering analyst Valentino Ricotta (from the cybersecurity research division of Thales), who was awarded a "bug bounty" of $20,000 (£15,000 ). He said: "What especially struck me with this device, that's been sitting on my bedside table for years, is that it's connected to the internet. It's constantly running because the battery lasts a long time and it has access to my Amazon account. It can even pay for books from the store with my credit card in a single click. Once an attacker gets a foothold inside a Kindle, it could access personal data, your credit card information, pivot to your local network or even to other devices that are registered with your Amazon account."

Ricotta discovered flaws in the Kindle software that scans and extracts information from audiobooks... He also identified a vulnerability in the onscreen keyboard. Through both of these, he tricked the Kindle into loading malicious code, which enabled him to take the user's Amazon session cookies — tokens that give access to the account. Ricotta said that people could be exposed to this type of hack if they "side-load" books on to the Kindle through non-Amazon stores.
Ricotta donated his bug bounties to charity...

The Wall Street Journal reports that 68% of CEOs "plan to spend even more on AI in 2026, according to an annual survey of more than 350 public-company CEOs from advisory firm Teneo." And yet "less than half of current AI projects had generated more in returns than they had cost, respondents said." They reported the most success using AI in marketing and customer service and challenges using it in higher-risk areas such as security, legal and human resources.

Teneo also surveyed about 400 institutional investors, of which 53% expect that AI initiatives would begin to deliver returns on investments within six months. That compares to the 84% of CEOs of large companies — those with revenue of $10 billion or more — who believe it will take more than six months.

Surprisingly, 67% of CEOs believe AI will increase their entry-level head count, while 58% believe AI will increase senior leadership head count.
All the surveyed CEOS were from public companies with revenue over $1 billion...

AI "has gutted entry-level roles in the tech industry," reports Rest of World.

One student at a high-ranking engineering college in India tells them that among his 400 classmates, "fewer than 25% have secured job offers... there's a sense of panic on the campus." Students at engineering colleges in India, China, Dubai, and Kenya are facing a "jobpocalypse" as artificial intelligence replaces humans in entry-level roles. Tasks once assigned to fresh graduates, such as debugging, testing, and routine software maintenance, are now increasingly automated. Over the last three years, the number of fresh graduates hired by big tech companies globally has declined by more than 50%, according to a report published by SignalFire, a San Francisco-based venture capital firm. Even though hiring rebounded slightly in 2024, only 7% of new hires were recent graduates. As many as 37% of managers said they'd rather use AI than hire a Gen Z employee...

Indian IT services companies have reduced entry-level roles by 20%-25% thanks to automation and AI, consulting firm EY said in a report last month. Job platforms like LinkedIn, Indeed, and Eures noted a 35% decline in junior tech positions across major EU countries during 2024...

"Five years ago, there was a real war for [coders and developers]. There was bidding to hire," and 90% of the hires were for off-the-shelf technical roles, or positions that utilize ready-made technology products rather than requiring in-house development, said Vahid Haghzare, director at IT hiring firm Silicon Valley Associates Recruitment in Dubai. Since the rise of AI, "it has dropped dramatically," he said. "I don't even think it's touching 5%. It's almost completely vanished." The company headhunts workers from multiple countries including China, Singapore, and the U.K... The current system, where a student commits three to five years to learn computer science and then looks for a job, is "not sustainable," Haghzare said. Students are "falling down a hole, and they don't know how to get out of it."

A former Chinese official who fled to the U.S. says Beijing has used advanced surveillance technology from U.S. companies to track, intimidate, and punish him and his family across borders. ABC News reports: Retired Chinese official Li Chuanliang was recuperating from cancer on a Korean resort island when he got an urgent call: Don't return to China, a friend warned. You're now a fugitive. Days later, a stranger snapped a photo of Li in a cafe. Terrified South Korea would send him back, Li fled, flew to the U.S. on a tourist visa and applied for asylum. But even there -- in New York, in California, deep in the Texas desert -- the Chinese government continued to hunt him down with the help of surveillance technology.

Li's communications were monitored, his assets seized and his movements followed in police databases. More than 40 friends and relatives -- including his pregnant daughter -- were identified and detained, even by tracking down their cab drivers through facial recognition software. Three former associates died in detention, and for months shadowy men Li believed to be Chinese operatives stalked him across continents, interviews and documents seen by The Associated Press show.

The Chinese government is using an increasingly powerful tool to cement its power at home and vastly amplify it abroad: Surveillance technology, much of it originating in the U.S., an AP investigation has found. Within China, this technology helped identify and punish almost 900,000 officials last year alone, nearly five times more than in 2012, according to state numbers. Beijing says it is cracking down on corruption, but critics charge that such technology is used in China and elsewhere to stifle dissent and exact retribution on perceived enemies.

Outside China, the same technology is being used to threaten wayward officials, along with dissidents and alleged criminals, under what authorities call Operations "Fox Hunt" and "Sky Net." The U.S. has criticized these overseas operations as a "threat" and an "affront to national sovereignty." More than 14,000 people, including some 3,000 officials, have been brought back to China from more than 120 countries through coercion, arrests and pressure on relatives, according to state information.

The pro-Russian CyberVolk group resurfaced with a Telegram-based ransomware-as-a-service platform, but fatally undermined its own operation by hardcoding master encryption keys in plaintext. The Register reports: First, the bad news: the CyberVolk 2.x (aka VolkLocker) ransomware-as-a-service operation that launched in late summer. It's run entirely through Telegram, which makes it very easy for affiliates that aren't that tech savvy to lock files and demand a ransom payment. CyberVolk's soldiers can use the platform's built-in automation to generate payloads, coordinate ransomware attacks, and manage their illicit business operations, conducting everything through Telegram.

But here's the good news: the ransomware slingers got sloppy when it came time to debug their code and hardcoded the master keys -- this same key encrypts all files on a victim's system -- into the executable files. This could allow victims to recover encrypted data without paying the extortion fee, according to SentinelOne senior threat researcher Jim Walter, who detailed the gang's resurgence and flawed code in a Thursday report.

Online retailer Coupang, often called South Korea's Amazon, is dealing with the fallout from a breach that exposed the personal information of more than 33 million accounts -- roughly two-thirds of the country's population -- after a former contractor allegedly used credentials that remained active months after his departure to access customer data through the company's overseas servers.

The breach began in June but went undetected until November 18, according to Coupang and investigators. Police have called it South Korea's worst-ever data breach. The compromised information includes names, phone numbers, email addresses and shipping addresses, though the company says login credentials, credit card numbers, and payment details were not affected.

Coupang's former CEO Park Dae-jun told a parliamentary hearing that the alleged perpetrator was a Chinese national who had worked on authentication tasks before his contract ended last December. Chief information security officer Brett Matthes testified that the individual had a "privileged role" giving him access to a private encryption key that allowed him to forge tokens to impersonate customers. Legislators say the key remained active after the employee left. The CEO of Coupang's South Korean subsidiary has resigned. Founder and chair Bom Kim has yet to personally apologize but has been summoned to a second parliamentary hearing.

joshuark shares a report from BleepingComputer: More than 10,000 Docker Hub container images expose data that should be protected, including live credentials to production systems, CI/CD databases, or LLM model keys. After scanning container images uploaded to Docker Hub in November, security researchers at threat intelligence company Flare found that 10,456 of them exposed one or more keys. The most frequent secrets were access tokens for various AI models (OpenAI, HuggingFace, Anthropic, Gemini, Groq). In total, the researchers found 4,000 such keys. "These multi-secret exposures represent critical risks, as they often provide full access to cloud environments, Git repositories, CI/CD systems, payment integrations, and other core infrastructure components," Flare notes. [...]

Additionally, they found hardcoded API tokens for AI services being hardcoded in Python application files, config.json files, YAML configs, GitHub tokens, and credentials for multiple internal environments. Some of the sensitive data was present in the manifest of Docker images, a file that provides details about the image.Flare notes that roughly 25% of developers who accidentally exposed secrets on Docker Hub realized the mistake and removed the leaked secret from the container or manifest file within 48 hours. However, in 75% of these cases, the leaked key was not revoked, meaning that anyone who stole it during the exposure period could still use it later to mount attacks.

Flare suggests that developers avoid storing secrets in container images, stop using static, long-lived credentials, and centralize their secrets management using a dedicated vault or secrets manager. Organizations should implement active scanning across the entire software development life cycle and revoke exposed secrets and invalidate old sessions immediately.

Google's Chrome team has built an experimental browser called Disco that takes a query or prompt, opens a cluster of related tabs, and then generates a custom application tailored to whatever task the user is trying to accomplish. The browser launched Thursday as an experiment in Google's Search Labs.

GenTabs, the core feature powering Disco, are information-rich pages created by Google's Gemini AI models -- ask for travel tips and the system builds a planner app; ask for study help and it creates a flashcard system. Disco -- named partly for fun and partly as shorthand for "discovery" -- started as a hackathon project inside Google before catching the team's imagination.

Parisa Tabriz, who leads the Chrome team, said that Disco is not intended as a general-purpose browser and is not an attempt to cannibalize Chrome. The experiment aims to test what happens when users move from simply having tabs to generating personalized, curated applications on demand. The capability relies on features in the recently launched Gemini 3, which can create one-off interactive interfaces and build miniature apps on the fly rather than just returning text or images.

Stanford researchers spent much of the past year building an AI bot called Artemis that scans networks for software vulnerabilities, and when they pitted it against ten professional penetration testers on the university's own engineering network, the bot outperformed nine of them. The experiment offers a window into how rapidly AI hacking tools have improved after years of underwhelming performance.

"We thought it would probably be below average," said Justin Lin, a Stanford cybersecurity researcher. Artemis found bugs at a fraction of human cost -- just under $60 per hour compared to the $2,000 to $2,500 per day that professional pen testers typically charge. But its performance wasn't flawless. About 18% of its bug reports were false positives, and it completely missed an obvious vulnerability on a webpage that most human testers caught. In one case, Artemis found a bug on an outdated page that didn't render in standard browsers; it used a command-line tool called Curl instead of Chrome or Firefox.

Dan Boneh, a Stanford computer science professor who advised the researchers, noted that vast amounts of software shipped without being vetted by LLMs could now be at risk. "We're in this moment of time where many actors can increase their productivity to find bugs at an extreme scale," said Jacob Klein, head of threat intelligence at Anthropic.

An anonymous reader quotes a report from Axios: OpenAI says the cyber capabilities of its frontier AI models are accelerating and warns Wednesday that upcoming models are likely to pose a "high" risk, according to a report shared first with Axios. The models' growing capabilities could significantly expand the number of people able to carry out cyberattacks. OpenAI said it has already seen a significant increase in capabilities in recent releases, particularly as models are able to operate longer autonomously, paving the way for brute force attacks.

The company notes that GPT-5 scored a 27% on a capture-the-flag exercise in August, GPT-5.1-Codex-Max was able to score 76% last month. "We expect that upcoming AI models will continue on this trajectory," the company says in the report. "In preparation, we are planning and evaluating as though each new model could reach 'high' levels of cybersecurity capability as measured by our Preparedness Framework." "High" is the second-highest level, below the "critical" level at which models are unsafe to be released publicly. "What I would explicitly call out as the forcing function for this is the model's ability to work for extended periods of time," said OpenAI's Fouad Matin.

New submitter emangwiro shares a report: The HDMI Forum, responsible for the HDMI specification, continues to stonewall open source. Valve's Steam Machine theoretically supports HDMI 2.1, but the mini-PC is software-limited to HDMI 2.0. As a result, more than 60 frames per second at 4K resolution are only possible with limitations. In a statement to Ars Technica, a Valve spokesperson confirmed that HDMI 2.1 support is "still a work-in-progress on the software side." "We've been working on trying to unblock things there."

The Steam Machine uses an AMD Ryzen APU with a Radeon graphics unit. Valve strictly adheres to open-source drivers, but the HDMI Forum is unwilling to disclose the 2.1 specification. According to Valve, they have validated the HDMI 2.1 hardware under Windows to ensure basic functionality.

Longtime Slashdot reader Randseed writes: With the likes of Google Nest, Ring, and others cooperating with law enforcement, I started to look for affordable wireless IP security cameras that I can put around my house. Unfortunately, it looks like almost every thing now incorporates some kind of cloud-based slop. All I really want is to put up some cameras, hook them up to my LAN, and install something like ZoneMinder. What are the most economical, wireless IP security cameras that I can set up with my server?