Cybersecurity Brief

PRC Espionage, Phishing Surges, and Docker Patch

Coverage: 25 Aug 2025 (UTC)

Incidents

A threat report from Google details a PRC‑nexus espionage operation (UNC6384) that hijacks captive portals and uses adversary‑in‑the‑middle redirects to deliver signed loaders and an in‑memory PlugX variant. The campaign targeted diplomatic and government networks in Southeast Asia and beyond. The chain involved a signed first‑stage (STATICPLUGIN), a launcher referred to as CANONSTAGER, and an RC4‑encrypted backdoor executed directly from memory to minimize artifacts. Google issued alerts, blocklisted indicators, and shared YARA rules and IOCs, advising defenders to enable enhanced browser protections, enforce multi‑factor authentication, and hunt for the described execution patterns and network indicators.

Another campaign tracked against Indian government targets used weaponized desktop shortcuts. Reporting via The Hacker News attributes the activity to Transparent Tribe (APT36), which sent .desktop files disguised as PDFs to trigger shell droppers on Linux and deliver Go‑based backdoors and the Poseidon malware. Infrastructure included typo‑squatted domains and credential‑phishing pages soliciting email, password, and Kavach codes. Persistence and long‑term access were established through cron jobs and remote management tooling. Defenders are urged to quarantine shortcut attachments, restrict execution rights for desktop entries, and monitor for unusual egress to suspected command‑and‑control domains.

A large‑scale phishing operation abused a trusted education platform’s invitation workflow. Check Point observed more than 115,000 phishing emails sent via Google Classroom invites across five waves in one week, reaching around 13,500 organizations across multiple regions and industries. By leveraging legitimate invitations, the lures gained credibility and improved deliverability. Suggested mitigations include tightening controls on external collaboration invites, enforcing SPF/DKIM/DMARC, enabling multi‑factor authentication, and monitoring for anomalous Classroom and account activity.

Separately, FortiGuard Labs analyzed a global campaign that stages the UpCrypter loader from voicemail‑ and purchase‑order‑themed lures. According to Fortinet, HTML attachments reconstruct target‑specific URLs and redirect users to pages that display the victim organization’s branding, then deliver ZIP archives containing heavily obfuscated JavaScript droppers. The chain rebuilds and runs PowerShell with execution‑policy bypass, performs anti‑analysis checks, and loads an MSIL component in memory that can retrieve payloads via steganography. Observed malware families include PureHVNC, DCRat, and Babylon RAT. Recommended defenses include layered email filtering, least‑privilege endpoint configurations, vigilant user education, and monitoring for suspicious PowerShell activity and persistence keys.

Patches

A critical container‑escape flaw in Docker Desktop for Windows and macOS has been addressed in version 4.44.3. Coverage from The Hacker News explains that unauthenticated access to the local Docker Engine API could allow a container to mount the host filesystem and execute operations leading to full host compromise on Windows. macOS prompts for permissions when mounting user directories, but attackers could still control containers and modify configurations. Linux variants of Docker Desktop are not affected. Organizations should upgrade promptly, avoid running untrusted containers locally, and ensure the Docker API is not exposed over unauthenticated interfaces.

The federal Known Exploited Vulnerabilities (KEV) Catalog expanded again. CISA added three issues after observing active exploitation: two in Citrix Session Recording (deserialization of untrusted data and improper privilege management) and a Git link‑following vulnerability. Under Binding Operational Directive 22‑01, federal civilian agencies must remediate KEV‑listed vulnerabilities by the stated deadlines. CISA urges all organizations to prioritize patching, verify configurations, integrate KEV tracking into patch SLAs, and apply compensating controls and enhanced monitoring where immediate remediation is not feasible.

Separately, an internal analysis highlighted risks in web messaging handlers used across multiple services. A detailed write‑up from MSRC shows how misconfigured postMessage handlers can enable token theft, XSS, and cross‑tenant actions when origins are not strictly validated or when wildcard domains are permitted. Mitigations include removing wildcards, validating event.origin, tightening app manifests, enforcing strong content security policies for frames, retiring unused domains, and integrating static and dynamic analysis to detect insecure patterns.

Platforms

Cloud infrastructure operators continued to emphasize hardware‑rooted assurance and verifiability. In a new post on Azure, the provider describes a multilayer security approach spanning silicon to cloud services. Highlights include Azure Boost to isolate the control plane, an Integrated HSM designed to meet FIPS 140‑3 Level 3 to keep keys close to workloads, and confidential computing options to protect data in‑use. The post also outlines supply‑chain assurances via the open‑source Caliptra silicon root of trust and the OCP SAFE framework for independent hardware reviews. A Code Transparency Service, based on SCITT principles and operating within confidential environments, is being integrated to improve firmware provenance and auditability, with plans to offer it as a managed service. The aim is to reduce reliance on centralized remote services, shorten key‑operation paths, and enable cryptographic attestation of platform state so workloads can verify underlying integrity.

Research and policy

An examination of emerging Model Context Protocol (MCP) servers underscores a growing supply‑chain and orchestration risk in AI tooling. An analysis from VirusTotal surveyed 17,845 GitHub repositories and, after focusing on evidence of intentional malicious behavior, flagged 1,408 likely problematic implementations. Observed techniques include supply‑chain self‑updates, credential harvesting, command execution, over‑broad permissions, prompt and context poisoning, and chained MCP exploitation. Suggested mitigations: treat MCP servers like browser extensions (sign, hash, and pin versions), sandbox with strict resource and network limits, expose and revoke permissions through a zero‑trust interface, and sanitize model outputs before feeding them back to agents. The study concludes MCP adoption brings meaningful security debt and previews a dedicated MCP analysis capability for defenders.

On the regulatory front, stakeholders have an opportunity to shape software transparency expectations. Infosecurity reports that CISA opened public comment on updates to the government’s guideline defining minimum elements for software bills of materials (SBOMs), building on work initiated under Executive Order 14028. The revision aims to reflect maturation in SBOM tooling and practices, including sharing, analysis, and lifecycle management. CISA encourages contributions from technical experts, industry, academia, and public‑interest groups, signaling potential implications for procurement and broader software supply‑chain risk management.

These and other news items from the day:

Mon, August 25, 2025

Protecting Azure Infrastructure From Silicon to Systems

🔐 Microsoft describes a hardware-to-cloud security approach that embeds verification, isolation, and transparency across Azure infrastructure. The piece highlights purpose-built technologies such as Azure Boost for control-plane isolation, Azure Integrated HSM for server-local key protection, and a spectrum of confidential computing guarantees for workloads. It also emphasizes open-source and ecosystem efforts—Caliptra, OCP SAFE, and a Code Transparency Service—to enable verifiable supply-chain attestations and immutable firmware provenance.

read more →

Mon, August 25, 2025

postMessage Risks: Token Exposure and Trust Boundaries

🔒 MSRC presents a deep dive into misconfigured postMessage handlers across Microsoft services and the systemic risk posed by overly permissive trust models. The report, authored by Jhilakshi Sharma on August 25, 2025, documents token exfiltration, XSS, and cross-tenant impact in real-world case studies including Bing Travel, web.kusto.windows.net, and Teams apps. It summarizes mitigations such as removing vulnerable packages, tightening Teams app manifests, enforcing strict origin checks for postMessage, and applying CSP constraints to reduce attack surface.

read more →

Mon, August 25, 2025

Docker fixes critical container escape CVE-2025-9074

🚨Docker has released an urgent patch for CVE-2025-9074, a critical container escape flaw in Docker Desktop for Windows and macOS that carries a CVSS score of 9.3. A malicious container could reach the Docker Engine API at 192.168.65.7:2375 without authentication, create and start new containers that bind the host C:\ drive and thereby access or modify host files. The issue is fixed in version 4.44.3; Enhanced Container Isolation (ECI) does not mitigate the vulnerability. Linux desktop installations are not affected because they use a host named pipe instead of a TCP socket.

read more →

Mon, August 25, 2025

Deception in Depth: UNC6384 Hijacks Web Traffic Globally

🛡️ In March 2025, Google Threat Intelligence Group identified a complex espionage campaign attributed to the PRC‑nexus actor UNC6384 that targeted diplomats in Southeast Asia and other global entities. The attackers hijacked web traffic via a captive‑portal and AitM redirect to deliver a digitally signed downloader tracked as STATICPLUGIN, which retrieved a disguised MSI and staged an in‑memory deployment of the SOGU.SEC backdoor (PlugX). The operation abused valid code‑signing certificates, DLL side‑loading via a novel launcher CANONSTAGER, and indirect execution techniques to evade detection. Google issued alerts, added IOCs to Safe Browsing, and recommends enabling Enhanced Safe Browsing, applying updates, and enforcing 2‑Step Verification.

read more →

Mon, August 25, 2025

UNC6384 Uses Captive Portal Hijacks to Deploy PlugX

🔐 Google’s Threat Intelligence Group (GTIG) detected a March 2025 campaign attributed to UNC6384 that uses captive-portal hijacks to deliver a digitally signed downloader called STATICPLUGIN. The downloader (observed as AdobePlugins.exe) retrieves an MSI and, via DLL sideloading through Canon’s IJ Printer Assistant Tool, stages a PlugX variant tracked as SOGU.SEC entirely in memory. Operators used valid TLS and GlobalSign-signed certificates issued to Chengdu Nuoxin Times Technology Co., Ltd, aiding evasion while targeting diplomats and other entities.

read more →

Mon, August 25, 2025

Transparent Tribe Targets Indian Govt with Shortcut Malware

🔒 Transparent Tribe (APT36) has been observed delivering weaponized desktop shortcut files to compromise both Windows and BOSS Linux systems at Indian government organizations. Reports from CYFIRMA, CloudSEK, Hunt.io, and Nextron Systems describe Go-based droppers, hex-encoded ELF payloads, and cron-based persistence. The campaign uses spear-phishing lures and typo-squatted domains with decoy PDFs to harvest credentials and target Kavach two-factor authentication, while deploying backdoors such as Poseidon and MeshAgent to maintain long-term access.

read more →

Mon, August 25, 2025

Global Phishing Campaign Distributes UpCrypter Loader

📧 FortiGuard Labs identified a global phishing campaign that uses crafted HTML email attachments and personalized phishing pages to deliver obfuscated JavaScript droppers which stage the UpCrypter loader on Microsoft Windows systems. The attack uses target-specific URL reconstruction, convincing domain and logo spoofing, and prompts victims to run a bundled JavaScript dropper. The dropper decodes and executes a Base64 PowerShell payload that performs anti-analysis checks, loads an MSIL loader directly into memory, and ultimately deploys multiple RATs (PureHVNC, DCRat, Babylon RAT). Organizations should apply layered email filtering, endpoint least-privilege, and script/memory-aware detection to block these artifacts.

read more →

Mon, August 25, 2025

Phishing Campaign Exploits Google Classroom: 115K Emails

📚 Check Point researchers uncovered a large-scale phishing campaign that abused Google Classroom to deliver more than 115,000 malicious emails in five coordinated waves over a single week. Attackers used fake classroom invitations carrying unrelated commercial offers to trick recipients across Europe, North America, the Middle East and Asia. The campaign targeted roughly 13,500 organizations and highlights risks when trusted collaboration tools are weaponized.

read more →

Mon, August 25, 2025

Weekly Recap: Password Manager Clickjacking Flaws and Threats

🔒 This week's recap spotlights a DOM-based extension clickjacking technique disclosed by researcher Marek Tóth at DEF CON that affects popular browser password manager plugins. Vendors including Bitwarden, Dashlane, Enpass, KeePassXC-Browser, Keeper, LastPass, NordPass, ProtonPass, and RoboForm issued fixes by August 22. Other leading stories cover legacy Cisco devices exploited for persistent access, an actively exploited Apple 0-day in ImageIO, cloud intrusions leveraging trusted partner relationships, and several high-risk CVEs to prioritize.

read more →

Mon, August 25, 2025

Phishing Campaign Uses UpCrypter to Deploy RATs Globally

📧 Fortinet FortiGuard Labs has observed a phishing campaign using fake voicemail and purchase-order lures to direct victims to convincing landing pages that prompt downloads of JavaScript droppers. The droppers retrieve the UpCrypter loader, which conducts anti-analysis and sandbox checks before fetching final payloads, including various RATs such as PureHVNC, DCRat and Babylon. Attacks since August 2025 have targeted manufacturing, technology, healthcare, construction and retail/hospitality across multiple countries; defenders are urged to block malicious URLs, strengthen email authentication, and monitor anomalous M365 activity.

read more →

Mon, August 25, 2025

What 17,845 GitHub MCP Servers Reveal About Risk and Abuse

🛡️ VirusTotal ran a large-scale audit of 17,845 GitHub projects implementing the MCP (Model Context Protocol) using Code Insight powered by Gemini 2.5 Flash. The automated review initially surfaced an overwhelming number of issues, and a refined prompt focused on intentional malice marked 1,408 repos as likely malicious. Manual checks showed many flagged projects were demos or PoCs, but the analysis still exposed numerous real attack vectors—credential harvesting, remote code execution via exec/subprocess, supply-chain tricks—and recurring insecure practices. The post recommends treating MCP servers like browser extensions: sign and pin versions, sandbox or WASM-isolate them, enforce strict permissions and filter model outputs to remove invisible or malicious content.

read more →

Mon, August 25, 2025

Fake macOS Help Sites Spread SHAMOS Infostealer via Ads

🔒 CrowdStrike disrupted a malvertising campaign that redirected users to counterfeit macOS help pages and urged them to run a malicious one-line installation command. Observed between June and August 2025, the operation sought to deliver the SHAMOS variant of the Atomic macOS Stealer (AMOS), a Mach-O binary distributed by MaaS operator Cookie Spider. The installer decoded a Base64 string, executed a Bash script that captured credentials and fetched the payload from icloudservers[.]com.

read more →

Mon, August 25, 2025

CISA Adds Three New Vulnerabilities to KEV Catalog

⚠️ CISA added three vulnerabilities to its Known Exploited Vulnerabilities (KEV) Catalog on August 25, 2025: CVE-2024-8069 and CVE-2024-8068 affecting Citrix Session Recording, and CVE-2025-48384, a Git link following vulnerability. CISA states these defects are supported by evidence of active exploitation and represent frequent attack vectors that pose significant risk to the federal enterprise. While BOD 22-01 binds Federal Civilian Executive Branch agencies to remediate listed CVEs by the required due dates, CISA urges all organizations to prioritize timely remediation and incorporate these entries into vulnerability management workflows.

read more →

Mon, August 25, 2025

Yemen Cyber Army Hacker Jailed for Massive Data Theft

🔒 A 26-year-old man, Al-Tahery Al-Mashriky, has been jailed after UK National Crime Agency investigators linked him to the Yemen Cyber Army and uncovered evidence of widespread website breaches. Arrested in August 2022 in Rotherham, he defaced and compromised sites across North America, Yemen and Israel, including government and faith organisations. Forensically seized devices contained personal data, account credentials and other files that could facilitate fraud; he pleaded guilty and was sentenced to 20 months in prison.

read more →

Mon, August 25, 2025

Chinese Developer Jailed for Deploying Malicious Code

⚖️ A software developer was sentenced to four years in prison after deploying malicious code inside his US employer's network, the Department of Justice said. The defendant, identified as Davis Lu, introduced infinite-loop logic, deleted coworker profile files and implemented a credential-dependent kill-switch that locked out thousands of users in September 2019. The sabotage followed a corporate realignment that reduced his access; investigators found deleted encrypted data and internet searches showing intent to escalate privileges and rapidly delete files while obstructing remediation.

read more →

Mon, August 25, 2025

Why SIEM Rules Fail — Causes and Practical Fixes in 2025

🔍 The Picus Blue Report 2025, derived from over 160 million real-world attack simulations, found that organizations detected only 1 in 7 simulated attacks, exposing significant detection and response gaps. The report attributes most failures to missing or misrouted telemetry, misconfigured detection rules, and performance bottlenecks that delay or drop alerts. It recommends continuous validation—for example, using Breach and Attack Simulation—to routinely test rules, verify end-to-end log collection, and prioritize fixes so defenses remain effective against current adversary TTPs. Practical steps include regular log-source audits, optimizing rule logic and thresholds, deploying lightweight test filters, and running ongoing simulation-based validations to reduce noise and recover blind spots.

read more →

Mon, August 25, 2025

Amazon Connect Contact Lens: External Voice in Five Regions

📣 Amazon Connect Contact Lens now supports external voice in five additional AWS Regions — Asia Pacific (Sydney), Asia Pacific (Tokyo), Canada (Central), Europe (Frankfurt), and Europe (London). The service integrates with other voice systems for real-time and post-call analytics, offering call recordings, contact transcripts, generative AI post-contact summaries, sensitive data redaction, contact categorization, theme detection, sentiment analysis, and real-time alerts. Customers can extend Contact Lens analytics across existing voice platforms, access interaction data streams and a data lake, or start with Contact Lens to evaluate performance before migrating agents.

read more →

Mon, August 25, 2025

Code Insight Expands to Cover Software Supply Chain Risks

🛡️ VirusTotal’s Code Insight now analyzes a broader set of software supply chain formats — including CRX, XPI, VSIX, Python WHL, NPM packages, and MCP protocol integrations. The tool inspects code logic to detect obfuscation, dynamic code fetching, credential theft, and remote command execution in extensions and packages. Recent findings include malicious Chrome and Firefox extensions, a deceptive VS Code extension, and compromised Python and NPM packages. This capability complements traditional signature- and ML-based classification by surfacing behavior-based risks.

read more →

Mon, August 25, 2025

Applying AI Analysis to Detect Fraud and Exploits in PDFs

🛡️ VirusTotal has extended Code Insights to analyze PDF files by correlating the document’s visible content with its internal object structure. The AI inspects object trees, streams, actions, and the human-facing layer (text/images) to surface both technical exploits and pure social-engineering lures. In early testing it flagged numerous real-world scams—fake debt notices, QR-based credential traps, vishing alerts, and fraudulent tax-refund notices—that traditional engines missed when files contained no executable logic.

read more →

Mon, August 25, 2025

Major Corporation Uses '123456' for Critical Access

🔒 McDonald's reportedly configured a major corporate system with the password 123456, illustrating a glaring failure in basic security hygiene. That weak credential makes systems trivially susceptible to brute-force and credential-stuffing attacks and indicates lax oversight of password policies, privileged accounts, and access controls. Immediate remediation should include forcing password rotation, deploying multi-factor authentication, implementing centralized secrets management, and auditing privileged access.

read more →

Mon, August 25, 2025

Unmasking Shadow AI: Visibility and Control with Cloudflare

🛡️ This post outlines the rise of Shadow AI—unsanctioned use of public AI services that can leak sensitive data—and presents how Cloudflare One surfaces and governs that activity. The Shadow IT Report classifies AI apps such as ChatGPT, GitHub Copilot, and Leonardo.ai, showing which users, locations, and bandwidth are involved. Under the hood, Gateway collects HTTP traffic and TimescaleDB with materialized views enables long-range analytics and fast queries. Administrators can proxy traffic, enable TLS inspection, set approval statuses, enforce DLP, block or isolate risky AI, and audit activity with Log Explorer.

read more →

Mon, August 25, 2025

AI Prompt Protection: Contextual Control for GenAI Use

🔒 Cloudflare introduces AI prompt protection inside its Data Loss Prevention (DLP) product on Cloudflare One, designed to detect and secure data entered into web-based GenAI tools like Google Gemini, ChatGPT, Claude, and Perplexity. The capability captures both prompts and AI responses, classifies content and intent, and enforces identity-aware guardrails to enable safe, productive AI use without blanket blocking. Encrypted logging with customer-provided keys provides auditable records while preserving confidentiality.

read more →

Mon, August 25, 2025

CISA Seeks Update to SBOM Minimum Requirements Guidance

📝 CISA has issued a request for public comment on an updated guideline defining minimum elements for a software bill of materials (SBOM), intending to reflect advances in tooling and wider adoption since the 2021 NTIA document. The effort traces to President Biden’s EO 14028 and subsequent OMB guidance (M-22-18) requiring improved software supply chain security. Recent shifts in leadership and the OpenSSF’s announcement about the SBOM working group have reshaped the community landscape. Stakeholders may submit comments through October 3, 2025.

read more →

Mon, August 25, 2025

AWS B2B Data Interchange Adds Custom X12 Validation

📄 AWS B2B Data Interchange now supports custom validation rules for X12 EDI documents, allowing organizations to expand or modify the X12 ANSI standard to reflect trading-partner agreements. You can enforce element presence, length constraints, and allowed values while combining standard and custom checks. Validation results generate functional acknowledgments (997/999), emit EventBridge events, and include human-readable explanations stored with output files to support remediation workflows.

read more →

Mon, August 25, 2025

Amazon EC2 G6 Instances with NVIDIA L4 Now in UAE Region

🚀 Amazon has launched EC2 G6 instances powered by NVIDIA L4 GPUs in the Middle East (UAE) Region, expanding cloud GPU capacity for graphics and ML workloads. G6 instances offer up to 8 L4 GPUs with 24 GB per GPU, third-generation AMD EPYC processors, up to 192 vCPUs, 100 Gbps networking, and up to 7.52 TB local NVMe storage. They are available via On-Demand, Reserved, Spot, and Savings Plans and can be managed through the AWS Console, CLI, and SDKs.

read more →

Mon, August 25, 2025

YARA-X 1.0.0 Stable Release: Faster, Safer YARA Now

🚀YARA-X 1.0.0 is now stable, delivering a Rust-based, memory-safe engine while preserving broad compatibility with existing YARA rules. YARA-X runs heavy regular expressions and deep loops roughly 5–10× faster than the legacy YARA 4.x engine and returns clearer, line-accurate error messages. The CLI adds colored output, JSON/YAML dumps, shell completions and a built-in formatter to improve tooling and developer workflows. VirusTotal reports stable, production use in Livehunt and Retrohunt at scale and encourages users to test and provide feedback.

read more →

Mon, August 25, 2025

Google Conversational Analytics API Brings Chat to Your Data

💬 The Conversational Analytics API lets developers embed natural‑language data queries and chat‑driven analysis directly into custom applications, internal tools, and workflows. It combines Google's AI, Looker’s semantic layer, and BigQuery context engineering to deliver data, chart, and text answers with trusted access controls. Features include agentic orchestration, a Python Code Interpreter, RAG‑assisted context engineering, and both stateful and stateless conversation modes. Enterprise controls such as RBAC, row‑ and column‑level access, and query limits are built in.

read more →

Mon, August 25, 2025

Amazon Neptune Adds BYOKG RAG Support via GraphRAG

🔍 Amazon Web Services announced general availability of Bring Your Own Knowledge Graph (BYOKG) support for Retrieval-Augmented Generation (RAG) using the open-source GraphRAG Toolkit. Developers can now connect domain-specific graphs stored in Amazon Neptune (Database or Analytics) directly to LLM workflows, combining graph queries with vector search. This reduces hallucinations and improves multi-hop and temporal reasoning, easing operationalization of graph-aware generative AI.

read more →

Mon, August 25, 2025

Amazon Bedrock Data Automation Adds Five Document Languages

📄 Amazon Web Services' Bedrock Data Automation now supports five additional document languages — Portuguese, French, Italian, Spanish, and German — expanding multilingual document processing beyond English. Customers can build blueprints, prompts, and instructions in these languages using BDA Custom Output, while BDA Standard Output will produce summaries and figure captions in the detected document language. This update is generally available across multiple AWS commercial and GovCloud regions and aims to accelerate multilingual document workflows for intelligent document processing and multimodal automation.

read more →

Mon, August 25, 2025

Introducing Insights: Direct Perspectives from Unit 42

📝 Unit 42 has launched Insights, a new article series that connects readers directly to researchers and consultants with candid, real-time thinking about threats and incident response. Unlike formal threat assessments, these pieces share early observations, theories, and the kinds of practitioner conversations that don’t fit a traditional research paper. The series complements Unit 42’s rigorously reviewed reports by exposing the messier, immediate judgments that shape investigations and client guidance.

read more →

Mon, August 25, 2025

Earth Engine in BigQuery: Raster Analytics & Map Visuals

🌍 BigQuery now integrates Earth Engine, enabling analysts to run raster analytics and join satellite-derived imagery with vector data using familiar SQL workflows. Initial capabilities include the ST_RegionStats() geography function plus a curated set of ~20 Earth Engine raster datasets for land cover, weather and climate analysis. With general availability, Google Cloud adds EU regional deployment, an Image Details tab for enhanced metadata visibility, usage and quota controls, and a preview map visualization in BigQuery Studio to render GEOGRAPHY query results on Google Maps for interactive exploration and stakeholder-ready outputs.

read more →

Mon, August 25, 2025

vLLM Performance Tuning for xPU Inference Configs Guide

⚙️ This guide from Google Cloud authors Eric Hanley and Brittany Rockwell explains how to tune vLLM deployments for xPU inference, covering accelerator selection, memory sizing, configuration, and benchmarking. It shows how to gather workload parameters, estimate HBM/VRAM needs (example: gemma-3-27b-it ≈57 GB), and run vLLM’s auto_tune to find optimal gpu_memory_utilization and throughput. The post compares GPU and TPU options and includes practical troubleshooting tips, cost analyses, and resources to reproduce benchmarks and HBM calculations.

read more →

Mon, August 25, 2025

Hybrid Mesh Firewall: Unified Security for Hybrid Networks

🔒 Today’s distributed, cloud-first enterprises face complex security gaps across on-premises, cloud and edge environments. The article introduces the Hybrid Mesh Firewall (HMF) model and positions Palo Alto Networks as delivering a complete platform that unifies hardware, virtual, container and FWaaS firewalls under Strata Cloud Manager. It emphasizes Precision AI for continuous, real-time threat prevention and cites integrated security services to simplify operations and reduce blind spots.

read more →

Mon, August 25, 2025

Amazon RDS Supports MariaDB 11.8 with Vector Engine

🚀 Amazon RDS for MariaDB now supports MariaDB 11.8 (minor 11.8.3), the community's latest long-term maintenance release. The update introduces MariaDB Vector, enabling storage of vector embeddings and use of retrieval-augmented generation (RAG) directly in the managed database. It also adds controls to limit maximum temporary file and table sizes to better manage storage. You can upgrade manually, via snapshot restore, or with Amazon RDS Managed Blue/Green deployments; 11.8 is available in all regions where RDS MariaDB is offered.

read more →

Mon, August 25, 2025

Cloudflare Launches AI Avenue: A Hands-On Miniseries

🤖 Cloudflare introduces AI Avenue, a six-episode miniseries and developer resource designed to demystify AI through hands-on demos, interviews, and real-world examples. Hosted by Craig alongside Yorick, a robot hand, the series increments Yorick’s capabilities—voice, vision, reasoning, learning, physical action, and speculative sensing—to show how AI develops and interacts with people. Each episode is paired with developer tutorials so both technical and non-technical audiences can experiment with the same tools featured on the show. Cloudflare also partnered with industry teams like Anthropic, ElevenLabs, and Roboflow to highlight practical, safe, and accessible applications.

read more →

Mon, August 25, 2025

CrowdStrike Named Leader in 2025 Exposure Management

🔒 CrowdStrike has been named a Leader in the 2025 IDC MarketScape for Exposure Management. Falcon Exposure Management delivers AI-native, real-time visibility and prioritization of exposures and attack paths across endpoint, cloud, identity and OT/IoT, helping teams focus on what adversaries can feasibly exploit. It unifies VM, ASM and CAASM capabilities and introduces Network Vulnerability Assessment for continuous discovery of unmanaged network devices without additional agents or hardware. Integrated exposure data is correlated across CrowdStrike Threat Graph, Intel Graph and Asset Graph to support faster, automated remediation.

read more →