In this week's news episode, we talk about Microslop banning its own name from Discord (Microslop can no longer be written), NVIDIA unlaunches drivers multiple times -- again, RAM and SSD prices taking down a game archive, and Sam Altman's new government partnership that he so gleefully boasted.

Nada acquired the CH170 Digital case from Deep Cool a few months ago and decided to use it to build a compact gaming PC on a limited budget.

The CERT Coordination Center has disclosed VU#431821, a severe command injection vulnerability (CVE-2026-2256) in ModelScope's MS-Agent framework, allowing attackers to execute arbitrary OS commands via crafted prompt inputs. The flaw affects the Shell tool, where unsanitized user-derived content bypasses fragile denylist filters through techniques like encoding or obfuscation. Successful exploitation grants full remote code execution (RCE) privileges matching the agent's process, risking system compromise, data theft, or persistence. Discovered by researcher Itamar Yochpaz, no official patch exists yet, prompting urgent mitigation advice including sandboxing, least-privilege execution, and restricting agents to trusted sources. Proof-of-concept code is publicly available, heightening the risk for users processing unvalidated external content.

South Korean tax authorities have suffered a major embarrassment after accidentally exposing a cryptocurrency wallet's full seed phrase in a public press release photo. The wallet contained approximately $4.8 million worth of seized PRTG tokens confiscated from tax evaders. By posting the sensitive information online, the officials effectively handed over access to the funds, resulting in their loss. The blunder has sparked widespread amusement and disbelief on social media. This incident highlights the ongoing risks of mishandling sensitive crypto data even among government agencies.

A persistent bug in Windows 11 versions 24H2 and 25H2 is causing in-place feature upgrades to wipe the contents of the Dot3Svc folder, which stores essential policies for IEEE 802.1X authentication on wired Ethernet connections. This results in affected systems losing the ability to connect to the internet or corporate networks via Ethernet, effectively "deleting" wired access post-upgrade.

The issue, first reported in earlier releases like 23H2 and still occurring in 2026 despite prior patches, primarily impacts enterprise and managed environments relying on secure wired authentication. Users can restore connectivity by temporarily connecting via Wi-Fi and running the command gpupdate /force in an elevated Command Prompt to reapply Group Policy settings.

Microsoft has not yet issued a permanent fix for this recurring problem in the latest builds.

KeygraphHQ has released Shannon, a fully autonomous open-source AI penetration testing tool that doesn't just scan for vulnerabilities-it actively exploits them, generating verifiable proof-of-concept attacks like SQL injections and authentication bypasses with zero false positives. Powered by advanced AI agents and browser automation, the framework achieved an impressive 96.15% success rate on the challenging XBOW benchmark and uncovered over 20 critical issues, including full database exfiltration, when tested against OWASP Juice Shop. Released under the AGPL-3.0 license, Shannon surged in popularity almost immediately, drawing widespread attention for bridging the security gap in fast-paced AI-assisted development environments. While hailed as a game-changer for continuous red teaming, experts caution about potential misuse without proper authorization and ethical guidelines. The tool is now available on GitHub, empowering developers to run comprehensive, autonomous security tests with a single command.

Google released its March 2026 Android Security Bulletin on March 3, addressing a record-breaking 129 vulnerabilities across the Android ecosystem, including critical remote code execution and elevation of privilege flaws.

Among the patches is a high-severity zero-day vulnerability (CVE-2026-21385) in the Qualcomm Display component, confirmed to be actively exploited in limited targeted attacks through memory corruption. The update splits into two patch levels-2026-03-01 for core platform fixes and 2026-03-05 for full protection, including hardware-specific issues-and applies to devices running recent Android versions, with Pixel phones receiving the rollout.

Google urges users to check their device's security patch level in settings and install the latest updates promptly to safeguard against potential exploitation. Source code patches are being published to the Android Open Source Project within 48 hours, while Google Play Protect continues blocking apps that attempt to exploit these issues.

Apple has launched the iPhone 17e as its latest budget-friendly smartphone, priced at $599 and now available for preorder with shipments starting March 11, 2026. Visually identical to the iPhone 16e, it features the same 6.1-inch Super Retina XDR OLED display and single 48MP rear camera but upgrades to the newer A19 processor, a faster custom C1X modem, doubled base storage starting at 256GB, and full MagSafe/Qi2 wireless charging support. Compared to the older iPhone SE (3rd gen), the 17e offers a significantly larger OLED screen, Face ID instead of Touch ID, up to 26 hours of video playback, and modern features like an Action Button. A new pink color option joins black and white, adding fresh appeal to the entry-level segment. While it retains a 60Hz refresh rate and single-lens camera, the upgrades make it a compelling step up from previous budget models.

NVIDIA has just released a new driver for its graphics cards. The NVIDIA GeForce 595.71 WHQL driver is optimized for Resident Evil Requiem and Marathon. This driver also fixes the fan control issues that were present in the previous driver that NVIDIA immediately took down.

NVIDIA GeForce 595.71 WHQL Driver Release Notes

Game Ready for Resident Evil Requiem

This new Game Ready Driver provides the best gaming experience for the latest new games supporting DLSS 4 technology including Resident Evil Requiem. In addition, there is Game Ready support for Marathon which features DLSS Super Resolution and NVIDIA Reflex.

Fixed Gaming Bugs

• The Ascent: Intermittent black bar on top of screen on GeForce RTX 50 Series GPUs [5859818]
• Total War: THREE KINGDOMS: Green artifacts appear on GeForce RTX 50 series [5745647]
• FINAL FANTASY XII The Zodiac Age crashes with fatal error after driver update [5741199]
• Call of Duty Modern Warfare (2019) displays image corruption after driver update [5733427]
• Quantum Break: Performance drops significantly on Act 4 Part 1 [5607678]

Fixed General Bugs

• 595.59: HW monitoring utilities not detecting all fans on the GPU [5934264]
• 595.59: One or more fans not spinning on GPUs after driver update [5934333]
• Blackmagic Design: AV1 decode crash with multiple obu in one packet [5671098]

Microsoft patched a critical zero-day vulnerability, CVE-2026-21513 (CVSS 8.8), in its February security update after confirming active exploitation targeting the MSHTML component. The flaw resided in ieframe.dll, allowing attackers to bypass the Mark-of-the-Web protection and Internet Explorer Enhanced Security Configuration, potentially leading to arbitrary code execution. Cybersecurity firm Akamai traced a malicious LNK file exploiting this issue to infrastructure linked with the Russian state-sponsored group APT28. Despite Internet Explorer's retirement, remnants like MSHTML continue to pose risks in legacy code paths. Organizations are urged to apply the patch immediately and monitor for related indicators of compromise.

Chinese tech giant Honor stunned attendees at Mobile World Congress 2026 in Barcelona by unveiling its groundbreaking Robot Phone, featuring a pop-out 200-megapixel camera mounted on a tiny 4DoF robotic gimbal. The innovative device enables AI-powered subject tracking, cinematic stabilization, gesture-based interactions, and even playful embodied AI movements like nodding or bobbing to music. Described by Honor as a "new species of smartphone," it combines advanced robotics with mobile imaging for more dynamic videography and user engagement. While full specs and global pricing remain under wraps, the company confirmed a launch in China later this year. The futuristic gadget has already sparked widespread excitement and debate about the next era of AI-integrated smartphones.

This fan-made 3D render depicts Sherry Birkin comforting Grace Ashcroft, the new FBI analyst protagonist from Resident Evil Requiem (RE9), in a dimly lit room with suggestive dialogue emphasizing warmth and intimacy, styled as revealing "bonus DLC" attire to humorously nod to the game's survival horror roots.

Resident Evil 9 Requiem - Bonus DLC pic.twitter.com/bT42RQeSX7

— Pristine Renders (@pristinerendrs) March 1, 2026

Microsoft has locked its official Copilot Discord server and hidden its chat history after users flooded it with variations of the derogatory term "Microslop," a viral meme mocking the company's low-quality AI integrations like Copilot in Windows. The backlash intensified when moderators initially attempted to block the word, only prompting creative workarounds that spiraled into widespread spam. This heavy-handed response has triggered the classic Streisand effect, amplifying the meme across social media rather than suppressing it. The once-inactive server, meant for Copilot discussions, became a focal point for user frustration with Microsoft's aggressive AI push. Critics say the incident highlights growing tensions between corporate control and online community sentiment in 2026.

Elie Habib, CEO of Anghami and a Lebanese tech entrepreneur, has launched World Monitor, an open-source AI-powered dashboard that provides real-time tracking of global conflicts, military movements, protests, infrastructure disruptions, and market signals. The tool aggregates data from dozens of live sources-including military flights, naval vessels, nuclear sites, and news feeds-into a unified browser-based interface resembling a CNN war room combined with a Bloomberg Terminal. Featuring AI-driven threat classification, instability indexes per country, and convergence alerts for overlapping signals, it has rapidly gained traction among analysts, journalists, and geopolitics enthusiasts. Released under the MIT license (with some variants noted as AGPL), the project is fully accessible on GitHub and runs live at worldmonitor.app. This "weekend hack" democratizes advanced OSINT capabilities previously limited to well-funded institutions.

A developer has created a prototype Android AI agent powered by Alibaba's Qwen3.5-27B model running on NVIDIA's DGX Spark hardware. The system uses a web-based UI to let the AI analyze phone screens, decide actions, and perform automated taps and interactions in near real-time. This setup delivers a remarkable 4x inference speedup compared to standard configurations, enabling smoother autonomous mobile operations. The breakthrough highlights China's rapid progress in open-source AI agents for practical applications like app automation and accessibility tools. Shared widely on X, the demo underscores growing global interest in efficient, local multimodal AI capabilities.

Someone has already built an Android agent using Qwen3.5-27B + DGX Spark, distributing tasks via a web UI so the model can read the screen, make decisions, and click automatically, resulting in a 4x speedup inference. 🤯 pic.twitter.com/RptsXypNkC

— China pulse 🇨🇳 (@Eng_china5) March 1, 2026

Anthropic's Claude AI app has rocketed to the #1 spot among free downloads on the Apple App Store in the US, dethroning OpenAI's ChatGPT which now sits at #2. The surge follows widespread user migration and boycott calls against ChatGPT after OpenAI partnered with the US Department of Defense (referred to by some as the "Department of War") to integrate it into the secure GenAI.mil platform on February 9, 2026. This military affiliation sparked ethical concerns among users wary of AI's role in defense applications, prompting many to switch to Claude amid the controversy. While OpenAI emphasized built-in safety guardrails and lawful-use restrictions in the deal, the backlash has fueled a temporary but dramatic shift in app rankings. Industry observers note that such political and ethical debates are increasingly influencing consumer choices in the competitive AI chatbot market.

Google has released the second beta version of Android 17 for developers. In doing so, it has revealed a list of new features that users can expect in the new OS. The first is Bubbles, which is a full multi-window capability, like a desktop OS. Although most useful on large-screen devices like foldables and tablets, it will work on regular phones as well. "Users can create an app bubble ... by long-pressing an app icon on the launcher." Also new is a Contacts Picker, which gives apps a more privacy-oriented way to gather specific contact data. It works similarly to the Photo Picker introduced in Android 13, in that it puts the OS in-between apps and contacts data, so that only minimum necessary contact data is shared with the app, instead of the app being given unrestricted access to all contacts data. A new "Handoff" API will enable cross-device app handoff, where users can start a task in an app on one device (such as an Android phone) and seamlessly continue the same task on another device (like an Android tablet). The API supports both native app-to-app handoffs, and app-to-web handoffs as a fallback. Android 17 will also support two new ranging technologies: UWB DL-TDOA for indoor navigation, and the Proximity Detection feature of the newest Wi-Fi standards. Another new privacy-enhancing API is the EyeDropper, which lets apps gather a color sample from any visible pixel without needing to "see" other apps. Finally, Android 17 introduces new restrictions on messaging APIs, intended to keep third-party apps from maliciously accessing one-time passcodes (OTP).

This analysis examines a surprisingly budget-friendly QD OLED display, focusing on its performance in both SDR and HDR modes. It highlights key trade-offs in build quality and technical specifications required to achieve this lower price point.

Security researchers uncovered exposed source code from OpenAI's identity verification provider Persona, revealing infrastructure that appears to link user selfies, biometrics, and personal details to U.S. government surveillance systems including references to ICE's ONYX tool* and FinCEN reporting.

The February 2026 findings sparked widespread privacy concerns, with claims that KYC data collected during ChatGPT access verification could enable broad monitoring and risk scoring across social media and financial records. A pro-China account amplified the story on X, warning users against sharing photos or sensitive information with ChatGPT, asserting it would be handed over for secret military use involving the U.S. and Israel.

OpenAI has partnered with the Department of Defense to deploy custom AI models on secure government networks, fueling speculation about deeper data ties despite no confirmed evidence of routine civilian chat handover. Persona and OpenAI have faced accusations but provided limited denials, leaving questions about the extent of government access to user verification data.

*ICE's ONYX is an AI-powered social media surveillance and open-source intelligence (OSINT) analysis tool developed by the company Fivecast, acquired by U.S. Immigration and Customs Enforcement (ICE) through a $4.2 million contract (as reported in early 2026 procurement details and EFF coverage from January 2026).The tool enables automated, continuous, and targeted collection of multimedia data from sources including major social media platforms, news streams, search engines, online marketplaces, and the dark web.ONYX constructs detailed "digital footprints" for individuals by aggregating biographical data across platforms, tracks shifts in sentiment and emotion in online activity, and assigns risk levels or threat assessments to people based on their digital behavior and connections.It forms part of ICE's broader expansion into internet surveillance capabilities amid increased funding for immigration enforcement tools, often used to generate leads without traditional warrants by leveraging publicly available or brokered data.Note that a separate February 2026 leak involving KYC provider Persona featured a government subdomain named "onyx.withpersona-gov.com," but Persona representatives stated this naming was coincidental (inspired by a Pokemon) and unrelated to Fivecast's ONYX tool used by ICE.

In a dramatic escalation, the Trump administration blacklisted AI firm Anthropic on February 27, 2026, designating it a "supply-chain risk to national security" after the company refused Pentagon demands to remove safeguards prohibiting mass domestic surveillance and fully autonomous lethal weapons. President Trump ordered all federal agencies to immediately cease using Anthropic's technology, granting the Pentagon a six-month phase-out period for existing integrations.

Hours later, rival OpenAI announced a deal with the Department of Defense to deploy its models on classified networks, incorporating similar safety restrictions on surveillance and autonomous weapons. The move has sparked accusations of selective enforcement, with Anthropic vowing legal action while experts question the implications for AI governance in national defense. This clash highlights deepening tensions between Silicon Valley's ethical guardrails and military priorities.

The video discusses the recent trends in RAM and GPU prices, observing that prices have peaked in late January 2026, largely due to decreased demand

A sophisticated scam impersonating a Zoom meeting tricked victims into visiting a fake website that simulated network issues and prompted a mandatory "update," silently downloading a malicious installer. The malware, a rogue version of the legitimate Teramind monitoring tool, deployed surveillance capabilities to log keystrokes, capture screenshots, track web activity, and more on compromised Windows machines. Discovered on February 11, 2026, via Microsoft Defender for Endpoint, the campaign infected 1,437 users globally within 12 days before wider reporting by Malwarebytes. Experts urge users to access Zoom only through official apps or direct URLs and to avoid clicking unexpected update prompts.

With slow DDR5 RAM prices going through the roof, and the expensive stuff through the stratosphere, we investigate if RAM speed matters in 2026

It's Friday, so here's something for you! :-D

Researchers from UC Riverside have disclosed AirSnitch, a new class of attacks that bypasses Wi-Fi client isolation, allowing malicious users on the same network to spoof identities, intercept traffic, and perform man-in-the-middle attacks—even on WPA3-secured setups. The vulnerability, presented at the NDSS 2026 symposium, affects popular consumer routers from brands like Netgear, TP-Link, Asus, D-Link, and Tenda, as well as open-source firmwares such as OpenWrt and enterprise environments, with every tested device proving susceptible to at least one attack variant. Root causes stem from improper management of broadcast encryption keys and unlinked MAC, IP, and key layers in Wi-Fi architecture, making simple patches insufficient. No fixes are currently available, underscoring the need for fundamental changes to Wi-Fi standards to secure public hotspots, homes, and offices.

Seattle Ultrasonics has launched the world's first ultrasonic chef's knife for home use, the C-200, which employs high-frequency vibrations-over 30,000 times per second-to drastically reduce cutting effort and prevent food from sticking to the blade.

The 8-inch Japanese steel blade, priced at $399, glides through ingredients like tomatoes, cheese, and dense vegetables with up to 50% less force, delivering cleaner cuts and easier cleanup.

Unveiled prominently at CES 2026 and now shipping to eager customers after quick sell-outs of initial batches, the innovative tool brings industrial-grade ultrasonic technology to everyday cooking. Reviews praise its game-changing performance, though some note the added cost of the charging system and battery considerations.

This vibrating wonder is redefining kitchen prep for home cooks seeking precision without traditional sharpening hassles.

Apple announced today that its iPhone and iPad have become the first and only unmodified consumer devices certified to handle classified information up to the NATO Restricted level. This milestone certification, achieved without any special software or settings after rigorous testing by the German government, applies to devices running iOS 26 and iPadOS 26 across all NATO nations. The approval highlights Apple's unmatched built-in security architecture and paves the way for broader adoption in government and defense sectors. No other consumer mobile device has met this stringent information assurance standard.

Kali Linux has introduced a groundbreaking integration with Anthropic's Claude AI, allowing penetration testers to execute complex security tools using natural-language prompts instead of manual terminal commands. Through the open-source Model Context Protocol (MCP), users can issue requests like "Port scan scanme.nmap.org and check for a security.txt file," with Claude interpreting them, running tools such as Nmap or Gobuster on a Kali environment, and delivering results autonomously. This setup bridges Claude Desktop on macOS or Windows with a Kali instance via a secure API, enhancing accessibility and efficiency in offensive security workflows while raising considerations for data privacy and new risks like prompt injection. The feature, detailed by the Offensive Security team in early 2026, marks a shift toward conversational, AI-augmented penetration testing.

Digital Foundry's John Linneman thinks Resident Evil Requiem might already be using PSSR 2.0 on PS5 Pro because of how significant and impressive the jump in results are and an additional embargo for the PS5 Pro that prevented them from using Pro only footage.

Freshly unveiled at Samsung's Unpacked event, the Galaxy S26 Ultra boasts superior raw specs compared to Apple's iPhone 17 Pro Max, including a sharper 1440x3120 QHD display, a massive 200MP main camera, blazing 60W fast charging, and innovative features like a Privacy Display. Leaked comparison graphics circulating on X highlight Samsung's edges in megapixels, resolution, and charging speed, while the iPhone counters with its optimized A19 Pro chip, seamless iOS ecosystem, and brighter 3000-nit screen. Online reactions split sharply, with Android fans praising the S26 Ultra's hardware firepower and Apple loyalists defending ecosystem lock-in and real-world polish. The perennial flagship rivalry heats up as consumers weigh raw power against refined experience in 2026's premium smartphone battle.

iPhone 17 Pro Max vs Galaxy S26 Ultra pic.twitter.com/fLu4QHvTdK

— Apple Hub (@theapplehub) February 26, 2026

The Galaxy S26 Ultra has a “ wow” feature with video Lock horizontally. pic.twitter.com/mLQi7iHBrE

— Ice Universe (@UniverseIce) February 26, 2026

Genuinely new hardware features in smartphones are rare these days, I appreciate this one pic.twitter.com/3saVJd6tEP

— Marques Brownlee (@MKBHD) February 26, 2026

Everyone thinks you need flashy RGB, "OC Edition" graphics cards, and $400 motherboards to build an amazing gaming PC. In this build, we're proving that you can stay in the SAME performance class as the overpriced, overhyped systems - without wasting money on shiny plastic, unnecessary features, or brand tax. No unicorn vomit. No "extreme" badges. Just smart parts selection and real-world performance.

Micron has labeled GPU memory capacity-particularly next-gen GDDR7-a critical "performance bottleneck" for gamers, arguing that higher VRAM dramatically cuts stuttering and pop-in.
The bold claim arrives amid persistent rumors that Nvidia has delayed or canceled its RTX 50 SUPER lineup, which was set to pack 50% more GDDR7 than base models.
These enhanced cards, like a potential 24GB RTX 5070 Ti SUPER, fell victim to AI-driven shortages crippling consumer GPU production.
Micron's fresh 24Gb GDDR7 modules at 36Gbps speeds highlight the irony, as supply woes persist into 2026.

The United Kingdom has introduced stricter regulations under the Media Act 2024, subjecting major streaming platforms like Netflix, Prime Video, and Disney+ to enhanced Ofcom oversight for video-on-demand services with over 500,000 UK users. R18-rated and mature content must now feature mandatory age verification, while unclassifiable material faces removal entirely. Platforms are required to block under-18s from accessing or even hearing restricted content to better safeguard children. The changes align streaming services more closely with traditional broadcasters' standards on harmful material protections, sparking debates over privacy and online freedom.

Samsung has announced several new AI features of its next flagship phones, which are expected to be announced this Wednesday. First "a new Galaxy camera experience" that will let users "turn a photo from day to night" and "restore missing parts of objects in images". It will also introduce new AI functions for capturing better low-light photos, and new ways to combine multiple photos into one. Samsung is also adding Perplexity as a new option for an AI agent users can ask questions of and give instructions to. Perplexity will be able to access Samsung Notes, Clock, Gallery, Reminder and Calendar, as well as "select third-party apps". Samsung's new flagship phones - presumably the Galaxy S26 lineup - will be announced this Wednesday during an event that starts at 1pm ET. Samsung is offering a variety of credits and discounts for those who "reserve" in advance of the event and/or pre-order a device.

WarChinese AI lab DeepSeek has withheld early access to its flagship V4 model from U.S. chip giants Nvidia and AMD, granting priority instead to domestic player Huawei for optimization. This break from industry norms prioritizes China's local hardware ecosystem amid ongoing U.S. export restrictions on advanced semiconductors. Reports indicate the model was trained on Nvidia's banned Blackwell chips, potentially violating controls, with DeepSeek possibly masking U.S. hardware traces. Experts suggest the move signals deepening U.S.-China tech decoupling with limited immediate impact on American firms.

Threat actors have created a convincing fake Avast website that impersonates the legitimate antivirus company's portal to target French-speaking users with a refund scam. The fraudulent site claims victims were wrongly charged €499.99 and pressures them to provide full credit card details-including number, expiry date, and CVV-for a supposed refund process. It employs sophisticated tactics like real-time JavaScript to insert the current date, live chat support via Tawk.to, and urgency warnings about time-limited cancellations to harvest personal and financial information. Security experts from Malwarebytes warn users to always access official sites directly and never enter card details for refunds, as legitimate vendors do not require them this way.

Microsoft has announced the impending end of support for Windows Server 2016 on January 12, 2027, along with Windows 10 Enterprise LTSB 2016 and Windows 10 IoT Enterprise LTSB 2016 on October 13, 2026, after which these versions will no longer receive security patches, bug fixes, or technical assistance. Organizations still relying on these legacy systems risk increased vulnerability to cyber threats once final updates are delivered. Microsoft urges customers to upgrade to modern alternatives like Windows Server 2025 or newer Windows 10/11 LTSC releases for continued security and compliance. Extended Security Updates (ESU) will be available for up to three years post-support as a paid, security-only bridge option.

Is your Hoover or Candy washing machine worth repairing, or is it designed to fail?

How Stealthy was the 7zip Malware?