Browser Guardrails, Fortinet SSO Bypass Fixes, and SAP Patches

AlphaEvolve on Google Cloud: Gemini-driven evolution

🔬 AlphaEvolve is a Gemini-powered coding agent on Google Cloud that automates evolutionary optimization of algorithms for complex, code-defined problems. It takes a problem specification, evaluation logic, and a compile-ready seed program, then uses Gemini models to propose mutated code variants and an evolutionary framework to select and refine the best candidates. Early internal results at Google demonstrate measurable efficiency improvements, and the AlphaEvolve Service API is available through a private Early Access Program for interested organizations.

Google Adds Layered Defenses to Chrome's Agentic AI

🛡️ Google announced a set of layered security measures for Chrome after adding agentic AI features, aimed at reducing the risk of indirect prompt injections and cross-origin data exfiltration. The centerpiece is a User Alignment Critic, a separate model that reviews and can veto proposed agent actions using only action metadata to avoid being poisoned by malicious page content. Chrome also enforces Agent Origin Sets via a gating function that classifies task-relevant origins into read-only and read-writable sets, requires gating approval before adding new origins, and pairs these controls with a prompt-injection classifier, Safe Browsing, on-device scam detection, user work logs, and explicit approval prompts for sensitive actions.

Fortinet warns of critical FortiCloud SSO bypass flaws

⚠️ Fortinet released patches for two critical FortiCloud SSO authentication bypass vulnerabilities (CVE-2025-59718, CVE-2025-59719) impacting FortiOS, FortiProxy, FortiSwitchManager, and FortiWeb. Attackers can abuse improper cryptographic signature verification in crafted SAML messages to bypass FortiCloud SSO controls. Administrators should disable FortiCloud SSO until devices are patched — either via System -> Settings in the GUI or with the provided CLI command — and apply the vendor firmware updates promptly. Fortinet also fixed related credential and password-hash issues (CVE-2025-59808, CVE-2025-64471).

Why AI Security Requires an Integrated Platform and Governance

🔒 Gartner and Palo Alto Networks argue that AI security must be treated as a platform problem to manage accelerating generative AI risk, cost and complexity. The post recommends a two‑phase path: start with AI usage control to govern third‑party GenAI consumption, then extend protections into AI application development and runtime. Prisma Browser, Prisma SASE and Prisma AIRS are presented as the integrated tooling to discover, govern and protect AI usage and models. Palo Alto highlights Unit 42, Huntr and autonomous red teaming as sources of continuous validation.

SAP patches three critical vulnerabilities in December

🔒 SAP released December security updates fixing 14 vulnerabilities across multiple products, including three critical flaws that could enable remote code execution and full system compromise. The most severe, CVE-2025-42880 (CVSS 9.9), is a code-injection issue in SAP Solution Manager ST 720. A Tomcat-related bundle tracked as CVE-2025-55754 (CVSS 9.6) affects SAP Commerce Cloud, and CVE-2025-42928 (CVSS 9.1) is a deserialization bug in SAP jConnect. Administrators are urged to deploy the provided fixes without delay.

Amazon EC2 C8gn Instances Expand to Ohio and UAE Regions

🚀 Amazon EC2 C8gn instances, powered by AWS Graviton4 processors, are now available in US East (Ohio) and Middle East (UAE). They deliver up to 30% better compute performance versus Graviton3-based C7gn instances, include 6th-generation Nitro Cards, and provide up to 600 Gbps of network bandwidth. C8gn supports sizes up to 48xlarge (up to 384 GiB memory), up to 60 Gbps EBS bandwidth, and Elastic Fabric Adapter (EFA) on select large sizes to improve cluster latency and throughput.

Amazon EC2 X8g Instances Now Available in Stockholm

🚀 These instances, powered by AWS Graviton4 processors, are now available in the Europe (Stockholm) region and provide up to 3 TiB of memory with increased memory per vCPU compared to prior Graviton4 instances. X8g targets memory-intensive workloads such as EDA, in-memory and relational databases, real-time analytics, and memory-heavy containerized applications. They offer larger sizes (up to 48xlarge), enhanced networking (up to 50 Gbps), EBS bandwidth up to 40 Gbps, and EFA/ENA Express support on larger sizes.

Microsoft Expands U.S. Cloud Infrastructure and Regions

☁️ Microsoft is expanding its U.S. cloud footprint with a new East US 3 region in the Greater Atlanta Metro, scheduled to open in early 2027, and by adding capacity and Availability Zones across multiple existing U.S. regions. The East US 3 region is designed for resilience with Availability Zones, support for advanced AI workloads, and sustainability goals including LEED Gold and water conservation. Microsoft is also increasing zone redundancy in North Central US, West Central US, and the US Government Arizona region to boost capacity, compliance, and mission readiness.

Nutanix NC2 Now Generally Available on Google Cloud

🚀 Nutanix Cloud Clusters (NC2) is now generally available on Google Cloud, enabling organizations to run their Nutanix hybrid cloud directly on Google Compute Engine bare metal without refactoring workloads. NC2 supports the Z3 and C4 machine families with high-density NVMe local SSDs, integrates Nutanix Flow virtual networking, and maintains unified management via Prism Central. The solution connects to Google data and AI services like BigQuery and Vertex AI, supports license portability, and will be purchasable through Google Cloud Marketplace.

Further Hardening of Mali GPU Drivers with SELinux

🔒 Google’s Android Security and Privacy team collaborated with Arm to analyze the Mali GPU driver and implement SELinux-based IOCTL filtering that reduces the kernel driver's attack surface. The team categorized IOCTLs as unprivileged, instrumentation, and restricted, and used a staged rollout—first opt-in testing via a gpu_harden attribute, then opt-out with a gpu_debug domain—to validate behavior in real devices. The post provides step-by-step guidance for vendors to adopt a platform-level macro, define device-specific IOCTL lists, and enforce policy to keep deprecated and debug IOCTLs unreachable in production.

Google deploys second model to guard Gemini Chrome agent

🛡️ Google has added a separate user alignment critic to its Gemini-powered Chrome browsing agent to vet and block proposed actions that do not match user intent. The critic is isolated from web content and sees only metadata about planned actions, providing feedback to the primary planning model when it rejects a step. Google also enforces origin sets to limit where the agent can read or act, requires confirmations for banking, medical, password use and purchases, and runs a classifier plus automated red‑teaming to detect prompt injection attempts during preview.

Shai-Hulud 2.0: Detecting and Defending Supply-Chain Attacks

🛡️ The Shai-Hulud 2.0 campaign is a widescale npm supply-chain compromise that injects malicious preinstall scripts to execute a bundled Bun runtime and harvest cloud credentials. Microsoft Defender observed attackers installing GitHub Actions runners named SHA1HULUD, using TruffleHog to locate secrets, and exfiltrating stolen credentials to public repositories. The guidance outlines detections, hunting queries, and prioritized mitigations for developers, maintainers, and cloud defenders.

Windows PowerShell Warns When Invoke-WebRequest Runs

⚠ Windows PowerShell 5.1 now displays a security confirmation when using Invoke-WebRequest to fetch web pages, warning that scripts in a downloaded page might run during parsing. The change, delivered with update KB5074204, mitigates a high-severity RCE tracked as CVE-2025-54100 and brings safer parsing behavior from PowerShell 7. Microsoft recommends rerunning commands with the -UseBasicParsing switch or updating automation to include it. Note that the 'curl' alias maps to Invoke-WebRequest and will trigger the same prompt.

California Man Pleads in $263M Cryptocurrency Theft

🔒 Evan Tangeman, 22, has pleaded guilty to laundering proceeds from a sophisticated criminal network that stole roughly US $263 million in cryptocurrency. Prosecutors say the Social Engineering Enterprise was organised via online gaming connections and used hackers, impersonating 'callers', burglars and money launderers to seize and convert victims' crypto. Tangeman admitted converting about US $3.5 million and faces sentencing on April 24, 2026.

React2Shell Exploits Deploy EtherRAT, Linked to DPRK

🔐 Security researchers at Sysdig report new campaigns exploiting React2Shell (CVE-2025-55182), resulting in a novel implant that delivers EtherRAT and demonstrates advanced persistence and evasion. The exploit targets React v19 and many related frameworks, using a base64 shell command to fetch a downloader that installs Node.js, decrypts an obfuscated JavaScript dropper, and executes a blockchain-based C2-capable payload. Sysdig observed tooling overlaps with North Korea-associated campaigns, though firm attribution remains unconfirmed.

Microsoft Patch Tuesday — December 2025 Security Fixes

🛡️ Microsoft released its final Patch Tuesday of 2025, addressing 56 vulnerabilities including one actively exploited zero-day, CVE-2025-62221, and two publicly disclosed bugs. The zero-day is a privilege escalation in the Windows Cloud Files Mini Filter Driver, a core component used by cloud sync services such as OneDrive. Three flaws received Microsoft’s Critical rating, including two Office bugs exploitable via Outlook’s Preview Pane. Administrators should prioritize updates for the flagged privilege escalation issues and apply patches promptly.

North Korea-linked Actors Use React2Shell to Deploy EtherRAT

🛡️ Threat actors tied to North Korea have been observed exploiting the critical React Server Components vulnerability (React2Shell, CVE-2025-55182) to deliver a new remote access trojan named EtherRAT. The implant downloads a Node.js runtime, decrypts and spawns a JavaScript payload, and resolves command-and-control via Ethereum smart contracts using a multi-endpoint consensus method. EtherRAT persists on Linux with five distinct mechanisms and supports self-updating obfuscated payloads, enabling long-term stealthy access and making remediation difficult.

North Korean Hackers Exploit React2Shell to Deploy EtherRAT

🔒 Researchers at Sysdig uncovered a new malware implant, EtherRAT, delivered via exploitation of the React2Shell deserialization flaw in Next.js just days after the vulnerability disclosure. The implant bundles a full Node.js runtime, uses an encrypted loader, and employs Ethereum smart contracts for resilient C2 while supporting five Linux persistence mechanisms. Operators can self-update the payload and execute arbitrary JavaScript, complicating detection and response.

IAB Abuses EDR and Windows Utilities for Stealthy Malware

🔐Storm-0249, an initial access broker, is abusing endpoint detection and response (EDR) components and trusted Windows utilities to execute malware stealthily. In one analyzed incident the actor used social engineering to run curl commands that installed a malicious MSI which drops a DLL placed beside the legitimate SentinelAgentWorker.exe, then performs DLL sideloading to run attacker code inside the signed EDR process. Additional payloads are piped into memory via PowerShell from a spoofed domain, avoiding disk-based detection. Researchers recommend behavior-based detection for trusted processes loading unsigned DLLs and stricter controls on curl, PowerShell, and living-off-the-land binaries.

CISA Adds Two Vulnerabilities to Known-Exploited Catalog

🔒 CISA has added two vulnerabilities to its Known Exploited Vulnerabilities (KEV) Catalog: CVE-2025-6218 (WinRAR path traversal) and CVE-2025-62221 (Microsoft Windows use-after-free). The agency cited evidence of active exploitation and emphasized that these flaws are frequent attack vectors posing significant risk to the federal enterprise. CISA reiterated that BOD 22-01 requires FCEB agencies to remediate cataloged CVEs by the required due dates and urged all organizations to prioritize timely remediation.

BYOVD Loader Used to Disable EDR in DeadLock Ransomware

🔐 Cisco Talos reported a novel Bring Your Own Vulnerable Driver (BYOVD) loader used to disable endpoint security and deliver DeadLock ransomware. The attacker exploited a Baidu Antivirus driver vulnerability (CVE-2024-51324) via a loader named EDRGay.exe and driver DriverGay.sys to terminate EDR processes at kernel level. A PowerShell payload bypassed UAC, disabled Windows Defender, stopped backup and database services, and removed all volume shadow copies. DeadLock uses a custom timing-based stream cipher and extensive kill and exclusion lists to encrypt files while avoiding system corruption.

Malicious VS Code Extensions and Supply‑Chain Packages

🔒 Security researchers uncovered malicious extensions on the Microsoft Visual Studio Code Marketplace that delivered stealer malware while posing as a dark theme and an AI assistant. Koi Security reported the extensions downloaded additional payloads, captured screenshots, and siphoned emails, Slack messages, Wi‑Fi passwords, clipboard contents and browser sessions to attacker servers. Microsoft removed the packages in early December 2025 after investigators linked them to a publisher using multiple similarly named packages.

December 2025 Patch Tuesday: One Zero-Day, 57 CVEs Addressed

🔔 Microsoft’s December 2025 Patch Tuesday addresses 57 CVEs, including one actively exploited Important zero‑day in the Windows Cloud Files Mini Filter Driver and two publicly disclosed Important zero‑days impacting GitHub Copilot for JetBrains and PowerShell. Two Critical RCE flaws in Microsoft Office increase urgency for enterprise patching and remediation. Organizations should prioritize applying Microsoft fixes, adopt layered mitigations where patches are delayed, and use CrowdStrike Falcon dashboards to track affected assets and remediation progress.

Microsoft issues KB5071546 ESU update for Windows 10

🔒 Microsoft has released the KB5071546 extended security update for Windows 10 Enterprise LTSC and systems enrolled in the ESU program, addressing 57 security vulnerabilities including three zero-days. The mandatory patch updates Windows 10 to build 19045.6691 (LTSC 2021 to 19044.6691) and installs automatically, requiring a restart. Notably, it fixes a remote code execution zero-day in PowerShell (CVE-2025-54100) by adding a confirmation prompt and guidance to use -UseBasicParsing with Invoke-WebRequest to avoid parsing embedded scripts.

Microsoft December 2025 Patch Tuesday: 57 Fixes, 3 Zero-Days

🔒 Microsoft's December 2025 Patch Tuesday delivers fixes for 57 vulnerabilities, including three zero-day flaws — one actively exploited and two publicly disclosed. The update addresses 19 remote code execution, 28 elevation of privilege, four information disclosure, three denial of service, and two spoofing issues across Windows, PowerShell, Office, Exchange Server and drivers. Administrators should prioritize the actively exploited CVE-2025-62221 and apply vendor patches promptly.

Ivanti warns of critical Endpoint Manager code flaw

⚠️ Ivanti is urging customers to patch a critical vulnerability (CVE-2025-10573) in its Endpoint Manager (EPM) that allows unauthenticated remote actors to execute arbitrary JavaScript via low-complexity cross-site scripting that requires user interaction. Reported by Rapid7, the flaw lets attackers join fake managed endpoints to poison administrator dashboards and hijack admin sessions when viewed. Ivanti released EPM 2024 SU4 SR1 and addressed three other high-severity bugs, while Shadowserver reports hundreds of Internet-facing EPM instances.

Spain Arrests 19-Year-Old Suspect Over 64M Data Records

🔒 A 19-year-old suspect in Igualada, Barcelona, was arrested after authorities linked him to breaches at nine companies and the theft of 64 million private records. Police say the dataset included full names, home addresses, email addresses, phone numbers, DNI numbers and IBAN codes that the suspect attempted to sell on hacker forums using multiple accounts and pseudonyms. Officers seized computers and cryptocurrency wallets believed to hold proceeds from the sales; the investigation began in June. Separately, Ukrainian police arrested a 22-year-old who used custom malware and a 5,000-account bot farm to compromise and sell social media access.

Streamlit Exposes Shadow AI Risks and Data Leaks at Scale

⚠️ UpGuard's analysis of Streamlit-hosted applications uncovered thousands of publicly accessible data apps that expose sensitive business and personal information. In October 2025 scans identified 14,995 unique IPs running Streamlit; after accounting for instances with authentication or errors, over ten thousand apps remained accessible without login. The report documents exposed PII and business intelligence dashboards and recommends practical controls: maintain an inventory of user apps, move sensitive workloads off the Community Cloud, and enable authentication by default.

DeadLock Ransomware Uses BYOVD to Disable Endpoint Defenses

🔒 Cisco Talos detailed a campaign where a financially motivated actor deployed DeadLock ransomware using a Bring Your Own Vulnerable Driver (BYOVD) technique to disable endpoint protections by exploiting a Baidu driver flaw (CVE-2024-51324). A custom loader invoked the vulnerable driver to issue kernel-level commands that killed security processes; PowerShell scripts then escalated privileges, stopped backup and security services, and erased shadow copies. The C++ payload (compiled July 2025) injects into rundll32.exe, uses a custom stream cipher with time-based keys to append ".dlock" and waits roughly 50 seconds to evade sandboxes; communications and ransom negotiations occurred via Session. Organizations should enforce MFA, maintain strong endpoint controls and keep regular offline backups.

Opportunistic Pro-Russia Hacktivists Attack Critical OT

🔒CISA, alongside the FBI, NSA, DOE, EPA, the Department of Defense Cyber Crime Center, and international partners, published a joint advisory describing opportunistic pro-Russia hacktivist activity targeting operational technology (OT) systems. These groups exploit minimally secured, internet-facing VNC connections to access OT control devices and have caused varying impacts, including physical damage. Named actors include Cyber Army of Russia Reborn, Z-Pentest, NoName057(16), and Sector16. The advisory recommends reducing internet exposure of OT assets, adopting mature asset-management and mapping practices, and enforcing robust authentication.

Pro-Russia Hacktivists Target Critical Infrastructure

⚠️ This joint advisory from CISA, FBI, NSA, and international partners details opportunistic intrusions by pro‑Russia hacktivist groups—CARR, NoName057(16), Z‑Pentest, and Sector16—against OT/ICS environments. Actors are exploiting internet‑exposed VNC services, using open‑source scanning and brute‑force tools to access HMI devices with default or weak credentials, causing loss of view, configuration changes, and operational downtime. The advisory urges organizations to reduce public exposure, apply network segmentation, enforce strong authentication (MFA where feasible), harden device credentials, and follow secure‑by‑design guidance for OT products.

CISA, FBI Warn: Protect Critical Infrastructure Now

🚨 CISA, the FBI, NSA, DOE, EPA, DOD’s DC3, and international partners issued a joint advisory alerting operators that pro‑Russia hacktivist groups are conducting opportunistic, low‑sophistication attacks against U.S. and global critical infrastructure. These actors exploit internet‑facing OT components (notably VNC and SCADA) and sometimes combine intrusions with DDoS. The advisory urges immediate mitigations: reduce OT exposure, improve asset management, and enforce robust authentication.

Critical Auth Bypass in India-Deployed CCTV Cameras

🔒 CISA reports a critical authentication bypass (CWE-306, CVE-2025-13607) affecting multiple India-deployed CCTV products, including D-Link DCS-F5614-L1. The flaw permits unauthenticated remote retrieval of device configuration and account credentials with low attack complexity and high impact. D-Link has released a software update for the DCS-F5614-L1; users should install the patch, verify firmware versions, and minimize network exposure while seeking guidance from other vendors.

STAC6565 Targets Canada; Gold Blade Deploys QWCrypt

🛡️ Sophos links nearly 40 intrusions from Feb 2024 to Aug 2025 to STAC6565, a cluster assessed to overlap the criminal group Gold Blade (aka RedCurl/Red Wolf). The campaign shows an unusually narrow geographic focus — almost 80% of attacks targeted Canadian organizations — and combines targeted data theft with selective ransomware deployment using QWCrypt. Attack chains abuse recruitment platforms to deliver multi‑stage loaders such as RedLoader and tools designed to evade AV and disable recovery, often leveraging WebDAV, Cloudflare Workers and program‑compatibility execution paths.

AMOS infostealer uses ChatGPT share to spread macOS malware

🛡️Kaspersky researchers uncovered a macOS campaign in which attackers used paid search ads to point victims to a public shared chat on ChatGPT that contained a fake installation guide for an “Atlas” browser. The guide instructs users to paste a single Terminal command that downloads a script from atlas-extension.com and requests system credentials. Executing it deploys the AMOS infostealer and a persistent backdoor that exfiltrates browser data, crypto wallets and files. Users should not run unsolicited commands and must use updated anti‑malware and careful verification before following online guides.

Ransomware Gangs Use Shanya Packer to Evade EDR Protections

🛡️ Shanya is a packer-as-a-service used by multiple ransomware gangs to conceal payloads that disable endpoint detection and response (EDR) tools. The service returns a custom, encrypted wrapper that decrypts and decompresses the payload entirely in memory and inserts it into a memory-mapped copy of shell32.dll, avoiding disk artifacts. Sophos telemetry links Shanya-packed samples to Medusa, Qilin, Crytox and Akira, and notes techniques that crash user-mode debuggers and facilitate DLL side-loading to deploy EDR killers.

Malicious VS Code Extensions Steal Credentials via DLL

🛡️ Researchers from Koi Security have uncovered two malicious Visual Studio Code extensions, Bitcoin Black and Codo AI, that delivered a DLL-based infostealer via a disguised Lightshot executable. The campaign used social engineering and evolving technical methods—initially complex PowerShell and passworded ZIPs, later streamlined to hidden batch scripts—to harvest screenshots, clipboard data, Wi‑Fi credentials and browser sessions. One extension posed as a theme while the other offered legitimate AI coding features, helping both evade suspicion on the VS Code Marketplace.

Storm-0249 Shifts to Fileless Execution and DLL Sideloader

🚨 ReliaQuest warns that Storm-0249 appears to be evolving from an initial access broker into an active operator, adopting domain spoofing, DLL side-loading and fileless PowerShell execution to facilitate ransomware intrusions. The actor used a Microsoft-mimicking URL and the Windows Run dialog to fetch and execute a PowerShell script that installed a trojanized SentinelOne DLL via a malicious MSI. This technique leverages living-off-the-land utilities and signed processes to maintain persistence and evade detection.

NCSC Warns Prompt Injection May Be Inherently Unfixable

⚠️ The UK National Cyber Security Centre (NCSC) warns that prompt injection vulnerabilities in large language models may never be fully mitigated, and defenders should instead focus on reducing impact and residual risk. NCSC technical director David C cautions against treating prompt injection like SQL injection, because LLMs do not distinguish between 'data' and 'instructions' and operate by token prediction. The NCSC recommends secure LLM design, marking data separately from instructions, restricting access to privileged tools, and enhanced monitoring to detect suspicious activity.

Experts Warn AI Is Becoming Integrated in Cyberattacks

🔍 Industry debate is heating up over AI’s role in the cyber threat chain, with some experts calling warnings exaggerated while many frontline practitioners report concrete AI-assisted attacks. Recent reports from Google and Anthropic document malware and espionage leveraging LLMs and agentic tools. CISOs are urged to balance fundamentals with rapid defenses and prepare boards for trade-offs.

U-Boot Bootloader: Improper Access to Volatile Boot Code

⚠️ U-Boot contains an improper access control vulnerability in volatile memory holding boot code (CVE-2025-24857) affecting all U-Boot versions prior to 2017.11 and several Qualcomm SoCs. Successful exploitation could allow arbitrary code execution; CISA reports a CVSS v4 base score of 8.6 with low attack complexity. Vendors advise upgrading to v2025.4, ensuring physical device security, and contacting Qualcomm support where appropriate.

Microsoft Patch Tuesday December 2025: 57 Vulnerabilities

🛡️ Microsoft released its December 2025 Patch Tuesday addressing 57 vulnerabilities, two labeled as critical and the remainder as important. Cisco Talos notes Microsoft assessed exploitation of the two critical issues as less likely, while several important flaws are considered more likely to be attacked. Talos published Snort and Snort 3 rules to detect exploitation attempts and recommends updating firewall SRUs and applying vendor patches promptly.

VMO2 and Google Cloud: Data Contracts for Scalable AI

🔒 VMO2, with Google Cloud, implemented data contracts as machine-readable agreements to guarantee dataset quality, schema, semantics, and SLOs for individual assets like BigQuery tables and Cloud Storage buckets. Defined in YAML and managed via GitLab, contracts are validated and operationalized by Dataplex Universal Catalog, which provisions Data Quality Scan jobs and profiling. The platform uses Cloud Composer, Pub/Sub, and BigQuery to orchestrate scans, surface results, and provide dashboards for real-time observability.

Windows 11 KB5072033 & KB5071417 Patch Tuesday December 2025

🔔 Microsoft released cumulative updates KB5072033 (25H2/24H2) and KB5071417 (23H2) as the December 2025 Patch Tuesday rollup. The mandatory updates include security fixes, bug patches, and new or enhanced features such as improved File Explorer dark mode, Virtual Workspaces advanced settings, and expanded Full‑Screen Experience for handheld devices. Install via Settings > Windows Update or the Microsoft Update Catalog; features will roll out gradually.

Partners Fuel Innovation with Cortex XSIAM & Prisma SASE

🚀 Palo Alto Networks announced that partners voted Cortex XSIAM as CRN’s 2025 Product of the Year for Security Operations Platform/SIEM and Prisma SASE as a 2025 Tech Innovator. Solution providers credited XSIAM’s AI-driven approach for sweeping the evaluation — leading in technology, revenue and customer need — and praised its ability to shift SOCs from tool management to outcome delivery. Partners highlighted Prisma SASE’s multicloud architecture, unified policies and AI copilot as essential for securing hybrid workforces, informed by feedback from over 70,000 customers and the recent Prisma SASE 4.0 release. Palo Alto frames these awards as validation of platform convergence and continued partner enablement.

Festo LX Appliance XSS Vulnerability (CVE-2021-23414)

⚠️ Festo SE & Co. KG's LX Appliance contains a cross-site scripting (XSS) vulnerability tied to the video.js library (CVE-2021-23414) that can allow crafted course content to execute scripts in high-privilege user sessions. The issue affects LX Appliance versions prior to June 2023 and has a CVSS v3.1 base score of 6.1. Festo coordinated disclosure with CERT@VDE and published advisory FSA-202301. Administrators should update affected appliances and apply recommended network isolation and secure remote access controls.

Gartner Urges Enterprises to Block AI Browsers Now

⚠️ Gartner has advised enterprises to block AI browsers until associated risks can be adequately managed. In its report Cybersecurity Must Block AI Browsers for Now, analysts warn that default settings prioritise user experience over security and list threats such as prompt injection, credential exposure and erroneous agent actions. Researchers and vendors have also flagged vulnerabilities and urged risk assessments and oversight.

Four clusters exploiting CastleLoader expand MaaS reach

🛡️Recorded Future's Insikt Group attributes rapid expansion of a modular loader ecosystem to an actor named GrayBravo, noting the distribution of a loader called CastleLoader under a malware-as-a-service model. The report identifies four distinct operational clusters that employ phishing, ClickFix campaigns, malvertising, and impersonation to deliver CastleLoader and secondary payloads such as CastleRAT and NetSupport RAT. These campaigns target logistics and enterprise software users and leverage multi-tiered C2 infrastructure and fraudulent platform accounts to increase credibility and resilience.

Amazon GameLift Servers Adds AI Assistance in Console

🤖 Amazon GameLift Servers now offers AI-powered assistance within the AWS Console, leveraging Amazon Q Developer to deliver tailored guidance for game developers. The integrated assistant helps with game server integration, fleet configuration, and performance optimization by surfacing in-console recommendations and troubleshooting steps. It is intended to streamline decision making, reduce troubleshooting time, and improve resource utilization for cost savings and better player experiences. The feature is available in all supported regions except AWS China.

Maintaining Enterprise IT Hygiene with Wazuh SIEM/XDR

🔒 Wazuh's IT hygiene capability delivers centralized, real-time inventory and configuration monitoring across all endpoints by leveraging the Syscollector module and dedicated indices. Security teams can quickly query hardware, OS, installed packages, running processes, user accounts, browser extensions, and open ports through an interactive dashboard. The feature supports detection of outdated software, unauthorized extensions, dormant or privileged accounts, and unexpected services, and it integrates with alerting and remediation workflows to enforce baselines and reduce attack surface.

AWS Partner Central Adds AI Deal Sizing for Opportunities

🔍 AWS Partner Central now includes AI-powered deal sizing within APN Customer Engagements (ACE) Opportunities, giving partners estimated monthly recurring revenue (MMR) and recommended AWS services when creating or updating opportunities. Partners can import AWS Pricing Calculator URLs to auto-populate service selections and spend estimates, with enhanced insights such as pricing optimization, cost-savings analysis, MAP eligibility, and modernization pathways. The feature is available worldwide via the console and the AWS Partner Central API for Selling.

Automating NIS2 Compliance: Move from Paperwork to Code

🛡️ The EU directive NIS2, in force in Germany since 06 December 2025, risks becoming a paperwork-heavy exercise unless organisations adopt automation and DevSecOps. The article argues security must be planned and enforced by technology, using Infrastructure as Code, policies-as-code and CI/CD pipelines so controls and evidence (commits, pipeline logs, SBOMs) are revision-proof. Solutions such as CIEM, CNAPP and SIEM can centralise IAM, vulnerability and incident data so auditability is produced by the platform rather than by post-hoc Word documents.

CISA Releases Three New Industrial Control Advisories

🔔 CISA published three Industrial Control Systems (ICS) advisories addressing vulnerabilities in Universal Boot Loader (U-Boot) (ICSA-25-343-01), the Festo LX Appliance (ICSA-25-343-02), and several India-based CCTV camera models (ICSA-25-343-03). Each advisory provides technical details, impact assessments, and recommended mitigations. CISA urges system operators, vendors, and administrators to review the advisories promptly and apply available updates or compensating controls to reduce operational risk.

Microsoft Investigates Copilot Outage Affecting Europe

⚠️Microsoft is mitigating an incident that has blocked or degraded access to its AI-powered Copilot service for users in the United Kingdom and parts of Europe. The company says telemetry points to an unexpected traffic surge that prevented service autoscaling, and engineers are manually scaling capacity to restore availability. A related admin-facing issue is also affecting some Microsoft Defender for Endpoint features.

Manufacturing Sees Fewer Encryptions but Ransom Risks

🔒 A recent Sophos study finds the manufacturing sector is blocking more ransomware before encryption, with only 40% of attacks resulting in data encryption this year versus 74% in 2024. Despite improved containment, data theft remains high (39% of encrypted cases) and more than half of affected firms paid ransoms; the median payment was about €861,000. Shortages of skilled staff, unknown vulnerabilities and inadequate protections are cited as root causes, and attacks are increasing stress and leadership pressures within IT teams.

Shifting Left at Enterprise Scale for Cloudflare Governance

🔐 Cloudflare describes how its Customer Zero team moved internal production account management from manual dashboard changes to a centralized Infrastructure as Code model to reduce human error and accelerate secure change. The effort uses Terraform, an Atlantis-driven CI/CD pipeline, and a custom tfstate-butler backend to securely manage state at scale. Policy enforcement relies on Open Policy Agent Rego policies executed through Conftest on every merge request, with warnings or deny gates and a formal exceptions workflow.

Changing the Physics of Cyber Defense with Graphs Today

🔍 John Lambert of MSTIC argues defenders should model infrastructure as directed graphs of credentials, entitlements, dependencies and logs so they can trace the attacker’s “red thread.” He introduces the algebras of defense—graphs, relational tables, anomalies, and vectors over time—that let analysts and AI ask domain-specific questions like blast radius or path to crown jewels. Lambert also emphasizes preventative hygiene: asset and entitlement management, deprecating legacy systems, segmentation, and phishing-resistant MFA. He urges collaborative intelligence and AI-enabled tooling to shift advantage back to defenders.

From Adoption to Impact — DORA AI Capabilities Model Guide

🤖 The 2025 DORA companion guide highlights that AI acts as an amplifier, boosting strengths and exposing weaknesses across teams. Drawing on a cluster analysis of nearly 5,000 technology professionals, it identifies seven foundational capabilities — including a clear AI stance, healthy and AI-accessible data, strong version control, small-batch workflows, user-centric focus, and quality internal platforms — that increase the odds of positive outcomes. The guide maps seven team archetypes to help leaders diagnose where to start and offers a Value Stream Mapping facilitation to direct efforts toward system-level constraints so AI-driven productivity scales safely.

Authorities Shut Down Thousands of Suspected Fraud Numbers

🔒 Investigators from the Cybercrime Center Baden-Württemberg, the Baden-Württemberg State Criminal Police (LKA) and BaFin said they have shut down more than 3,500 phone numbers believed to be used by investment, grandchild and fake-police scammers. The affected landline, mobile and VoIP numbers were disabled by providers, and about 350 Austrian numbers were taken offline in coordination with Vienna. The measure is part of Operation Herakles, intended to dismantle the technical infrastructure of fraud networks and protect consumers.

Hidden Forensic Evidence in Windows ETL: Diagtrack File

🔍 FortiGuard IR analysts discovered that an obscure ETL file, AutoLogger-Diagtrack-Listener.etl, can retain historical process execution data useful for post-incident forensics. Parsing ETW payloads exposed ProcessStarted events including ImageName, ProcessID, ParentProcessID and sometimes CommandLine entries that revealed deleted tools. Controlled testing showed creating the autologger and setting AllowTelemetry=3 often produced an empty file, indicating the DiagTrack service may populate the file only under undocumented conditions. Further research is needed to understand when and how this telemetry is written.

40,000 Phishing Emails Masquerade as E‑Signing Services

📧Attackers impersonating file-sharing and e-signature platforms sent over 40,000 finance-themed phishing emails, researchers at Check Point report. These messages mimicked notifications from services like SharePoint and popular e-signing vendors to coax recipients into clicking links or entering credentials. The campaign targeted finance workflows and aimed to harvest credentials or deliver follow-on malware, underscoring the need for robust email security and user vigilance.

Racks, Sprawl and the Myth of Redundancy in Modern Networks

🔁 The article traces redundancy from tangible rack-level practices to fragile cloud and software-defined environments. It argues that physical diversity, disciplined configuration management and automation remain essential as networks span BGP, SD-WAN, edge devices and cloud control planes. Real resilience requires policy alignment, diverse DNS and routing protections and rehearsed pre-mortems so backups are usable when they matter most.

Streamlining Zero Trust with a Shared Signals Framework

🔐 This guide shows how to operationalize the Shared Signals Framework (SSF) to deliver continuous device posture signals into identity platforms. It details a proof‑of‑concept workflow using Tines to receive webhooks from Kolide, enrich and map device data, generate and sign Security Event Tokens (SETs), and forward them to Okta as CAEP events. The approach enables real‑time policy enforcement and simplifies SSF adoption when endpoints lack native support. Steps and required credentials are summarized for quick deployment.

Whaling attacks against executives: risks and mitigation

🎯 Whaling attacks are highly targeted social engineering campaigns aimed at senior executives that combine reconnaissance, spoofing, and urgency to trick leaders into divulging credentials, approving transfers, or executing malware-laden actions. Threat actors exploit executives’ visibility, limited time, and privileged access, and increasingly leverage generative AI and deepfakes to scale and refine impersonations. Key defenses include personalised executive simulations, strict multi-party approval flows for high-value transfers, AI-enhanced email filtering, deepfake detection, and a Zero Trust approach to access.

AI vs Human Drivers — Safety, Trials, and Policy Debate

🚗 Bruce Schneier frames a public-policy dilemma: a neurosurgeon writing in the New York Times calls driverless cars a “public health breakthrough,” citing more than 39,000 US traffic fatalities and thousands of daily crash victims, while the authors of Driving Intelligence: The Green Book argue that ongoing autonomous-vehicle (AV) trials have produced deaths and should be halted and forensically reviewed. Schneier cites a 2016 paper, Driving to safety, which shows that proving AV safety by miles-driven alone would require hundreds of millions to billions of miles, making direct statistical comparison impractical. The paper argues regulators and developers must adopt alternative evidence methods and adaptive regulation because uncertainty about AV safety will persist.

AWS: Tagging for RDS and Aurora Automated Backups Released

🔖 Amazon Web Services now supports resource tagging for automated backups and cluster automated backups in Amazon RDS and Aurora. You can tag automated backups independently from the parent DB instance or DB cluster using the AWS Management Console, API, or SDK. Use these tags with IAM policies to implement attribute-based access control and to organize, manage, and track backup costs. This capability is available in all AWS Regions, including AWS GovCloud (US).

The AI Fix #80: DeepSeek, Antigravity, and Rude AI

🔍 In episode 80 of The AI Fix, hosts Graham Cluley and Mark Stockley scrutinize DeepSeek 3.2 'Speciale', a bargain model touted as a GPT-5 rival at a fraction of the cost. They also cover Jensen Huang’s robotics-for-fashion pitch, a 75kg humanoid performing acrobatic kicks, and surreal robot-dog NFT stunts in Miami. Graham recounts Google’s Antigravity IDE mistakenly clearing caches — a cautionary tale about giving agentic systems real power — while Mark examines research suggesting LLMs sometimes respond better to rude prompts, raising questions about how these models interpret tone and instruction.