AI Safety Guidelines: 10 Essential Rules to Protect Data, Finances, and Reputation When Working with LLMs
A detailed analysis of emerging AI-related security risks highlights how large language models can autonomously execute attack chains, fall victim to prompt injection, and cause cascading errors in complex workflows. The article examines real-world incidents such as the Anthropic vending machine pricing failure, the Meta Instagram account takeover via overly helpful AI support, and Copilot Studio data leaks through prompt injection. It emphasizes that while attack methods themselves are not revolutionary, AI agents can now scale them at machine speed with autonomous decision-making and recovery capabilities. The piece provides ten concrete safety rules covering financial controls, fact verification, data confidentiality, context pollution prevention, and access limitation. It also stresses that ultimate responsibility always remains with the human operator, not the AI system.
Translated from Russian
Read full articleLatest News
InfotecsTech Builds Custom Kubernetes-Based Traffic Generator for NGFW RnD and Performance Testing
InfotecsTech developed an in-house traffic generator to support development and testing of its high-performance NGFW cluster in active-active mode. The team rejected commercial solutions from IXIA and Xinertel due to high cost, insufficient flexibility for complex NGFW functions, and geopolitical restrictions. The resulting platform runs on Kubernetes with a master node managing Registry, Discovery, and Crux components while worker nodes host containerized generators. Supported generators include Cisco TRex for throughput and connection testing, SIPp for VoIP scenarios, pyftpdlib-based FTP generator, Yandex Tank with Nginx for live TLS traffic, and Selenium-based legitimate clients against OWASP Juice Shop. Practical scenarios cover VoIP call storms, maximum concurrent connections, 400 Gbit/s UDP throughput, 5 million CPS, and IMIX traffic at 300 Gbit/s with packet loss analysis. The system integrates Camunda for full automation of test scenarios and device configuration.
Translated from Russian
Kaspersky MDR Adds Automatic Correlation with Leaked Credentials via Digital Footprint Intelligence
Kaspersky has updated its Managed Detection and Response service to automatically match security events against data from compromised logins and passwords. The enhancement integrates Kaspersky Digital Footprint Intelligence to provide analysts with additional context when suspicious activity coincides with known credential leaks. According to the company, a quarter of attacks investigated in 2025 began with the use of stolen credentials. The update also introduces notifications for asset protection status, allowing administrators to address connectivity or telemetry issues that could affect monitoring quality. Managed service providers can now configure per-client license usage limits, and the service adds support for Kaspersky Embedded Systems Security for Linux 4.0. The MDR platform continues to deliver 24/7 infrastructure monitoring, threat hunting, incident investigation, and response capabilities.
Translated from Russian
Why AI Chatbots Misread Polished Reports and How to Prepare AI-Ready Content
Beautifully designed reports often confuse AI systems because visual layout does not preserve logical relationships between elements. When design is stripped away, machines may lose connections between headings, numbers, tables, and footnotes, leading to incorrect interpretations of key facts such as revenue growth. The solution is to create AI Ready content that maintains structure, semantics, and context even after text extraction or copying. This approach aligns closely with web accessibility standards from W3C and benefits both human readers using assistive technologies and automated analysis tools. Organizations are advised to use tagged PDFs following PDF/UA and ISO 14289-2:2024, provide data in XLSX or CSV alongside visual charts, and ensure every important figure travels with its full context including period, unit, and comparison base. The same principles apply to presentations, press releases, websites, and multimedia content.
Translated from Russian
Russia's Data Leak Penalties: 2.6 Million Rubles in Fines Despite 1.58 Billion Records Exposed in 2025
Russia introduced turnover-based fines for personal data leaks through Federal Law 420-FZ in late 2024, fundamentally altering the economics of information security investments. Over the first 18 months, Roskomnadzor opened 52 administrative investigations and issued 40 protocols totaling just 2.6 million rubles in penalties, with zero turnover fines applied. This occurred against a backdrop of 1.58 billion compromised records in 2025 alone. Public data leaks dropped fourfold in the first half of 2026, yet trading activity on underground forums rose nearly 60 percent as operators shifted to private sales. The law now ties penalties directly to the number of affected individuals and adds a turnover component for repeat violations under Article 13.11 of the Code of Administrative Offenses. Analysts note that the mere threat of larger fines has prompted companies to reassess data retention policies and risk models even without actual enforcement precedents.
Translated from Russian
Statistical Analysis in OSINT: Tools, Methods and Real-World Intelligence Applications
The article explains why statistical processing has become an essential component of professional OSINT work when analysts face large volumes of raw data. It outlines five core tasks that statistical methods solve: actor profiling, disinformation monitoring through time-series and graph analysis, financial intelligence, geospatial verification, and threat assessment. The text compares popular tools including Maltego, Gephi, Python, R, Power BI, Tableau and SpiderFoot, stressing that real investigations usually combine several of them. Detailed examples show how Pearson correlation, Louvain clustering and TF-IDF or BERT embeddings help identify coordinated botnets in social media. Famous leaks such as Panama Papers and Pandora Papers are presented as landmark cases where regression models, cluster analysis and network graphs exposed hidden ownership structures. The piece also lists open statistical sources from national agencies, international organisations and technical platforms, and reviews key mathematical techniques from basic descriptive statistics to ARIMA, CUSUM, DBSCAN and dimensionality reduction methods.
Translated from Russian
Russia Authorizes Temporary State Takeover of Unprotected Critical Infrastructure
President Vladimir Putin has signed a decree that empowers the Russian government to appoint temporary managers for critical infrastructure facilities whose owners have failed to ensure adequate security. The measure directly targets operators of objects classified as critical infrastructure who have not met protection requirements. Under the new rules, the state can intervene by installing an interim administrator to oversee operations until security standards are satisfied. This approach aims to prevent potential disruptions or threats arising from insufficiently defended assets. The decree provides a legal mechanism for rapid governmental response without permanent nationalization of the facilities. It reflects ongoing efforts to strengthen oversight of sectors deemed essential to national security and stability.
Translated from Russian
Windows 11 May Silently Remove Discrete GPU Drivers Due to Eco Mode on Gaming Laptops
Windows 11 has introduced an unexpected behavior that can delete drivers for discrete graphics cards on gaming laptops when the device remains powered off for an extended period in Eco Mode. The issue was first reported by the owner of an Asus ROG Zephyrus G14 equipped with a mobile GeForce RTX 5070 Ti, where the system treated the physically disconnected GPU as permanently removed after 15 days. Windows component pnpclean.dll then purged both the device entry and the associated Nvidia driver package, leaving the laptop with only a Microsoft Basic Display Adapter reporting error code 10. The default cleanup timer is normally 30 days, but it can be shorter for certain hardware configurations that fully disconnect discrete graphics for power saving. Similar incidents have been observed on older Asus ROG models with AMD Radeon GPUs, prompting the developer of G-Helper to add a warning to the utility. Users can mitigate the problem by periodically switching to Standard Mode or by disabling automatic driver package cleanup via a specific PowerShell command that sets the Autorun value to 0 under the Device Driver Packages registry key. Microsoft has not yet issued an official statement on the matter.
Translated from Russian
Bot Traffic Overtakes Human Traffic in 2024 as AI Agents and Scrapers Surge
Analysis of internet traffic from 2013 to 2026 shows automated bots steadily eroding human dominance online. Imperva data reveals human traffic fell to 47 percent by 2025 while malicious bots reached 40 percent. Good bots such as search crawlers remain stable, but gray AI agents and scrapers now drive much of the growth. Companies face rising infrastructure costs from bot traffic that generates no revenue, described as an invisible tax. Cloudflare and Akamai reports confirm high volumes of automated requests, with many classified as harmful scraping. The trend raises concerns about a synthetic internet shaped more by AI recommendations than human activity.
Translated from Russian
Understanding Security Alerts Only Becomes Clear When You Start Writing Detection Rules Yourself
A cybersecurity professional who spent years as a SOC analyst reviewing alerts before moving into detection engineering shares key insights on why alerts behave differently depending on perspective. The article explains the inherent trade-offs in tuning detection rules, the architectural reasons alerts often arrive without full context, and why default priorities rarely match any single organization's environment. It highlights how false positives are not simply developer errors but calculated risks between coverage and noise. The piece also stresses the value of feedback loops between SOC teams and detection developers to improve overall security effectiveness.
Translated from Russian
From Russian sources
Translated from Russian
Microsoft Removes Reinstallation Requirement for Enabling Smart App Control in Windows 11
Microsoft has eliminated the primary restriction on Smart App Control, allowing home users to activate or reactivate the Windows 11 security feature without performing a clean installation of the operating system. Previously, toggling the setting effectively required users to reinstall the entire system. Smart App Control leverages Microsoft's cloud-based reputation system and analyzes digital signatures to block suspicious, potentially dangerous, or unsigned files before execution. The toggle is now accessible through Windows Security under App & browser control. The update is being rolled out gradually via Windows 11 updates, though users who have disabled optional diagnostic data may still need a reset or reinstallation. The feature can interfere with developers and enterprise users working with rare or unsigned tools, as there is no option to whitelist individual blocked applications.
Russian State Operators Must Report Cyber Incidents to FSB Within 24 Hours From September 2026
Starting 1 September 2026, operators of state information systems in Russia will be required to notify the National Coordination Center for Computer Incidents within 24 hours of detecting a cyber incident. The mandate is set out in FSB Order 297 and applies to operators of GIS, information systems of state bodies, state unitary enterprises and state institutions. The order places particular emphasis on incidents involving unauthorized data transfers. Organizations will interact with GosSOPKA through the technical infrastructure of NKTsKI, which will issue an incident identifier upon receipt of the report. The new rules formalize what was previously considered good practice and stem directly from Law 568-FZ adopted in December 2025. The requirement removes any possibility of delaying notification to the FSB.
Telegram Desktop Adds WEB Proxy to Hide MTProxy Traffic Inside HTTPS and WebSocket Connections
Telegram Desktop has received an initial implementation of a new WEB proxy designed to conceal messenger connections within ordinary HTTPS and WebSocket traffic. The feature makes user activity appear as simple website browsing to network providers while MTProxy continues to operate underneath. Developers inserted more than 3,000 lines of code, including a dedicated WEB-proxy transport, domain configuration options, and interface elements such as “Open browser” and “Waiting for browser.” The mechanism works by opening an embedded web page that establishes a secure WebSocket link to a domain resembling a regular site. Multiple MTProxy connections are multiplexed into a single stream and sent through this channel. On the server side an intermediate node demultiplexes the stream and forwards the individual connections to a standard MTProxy instance without decrypting message content.
Trusting Russian Root Certificates and Monitoring Domestic CT Logs
The article examines risks associated with installing Russian root certificates issued by the Ministry of Digital Development. It explains how these certificates, when trusted, enable potential MitM attacks through TSPU infrastructure by allowing on-the-fly issuance of fraudulent certificates for foreign domains. Yandex Browser stands out by enforcing Certificate Transparency checks for domestic certificates, unlike other browsers that disable CT validation in the presence of added roots. Three primary domestic CT logs are maintained by Yandex, VK, and the Ministry, with log lists updated annually. A Python script is provided to query these logs directly and verify SCT inclusion for any certificate. The piece also notes limitations of existing web monitors such as ct.tlscc.ru when dealing with newer log endpoints.
Hermes Emerges as Modular Harness for Practical AI Security Testing
The article details the development of Hermes, a lightweight modular harness designed to turn raw language models into reliable AI security agents. Unlike heavy frameworks such as OpenClaw, Hermes uses dynamic skill loading, strict runbooks stored in Git, and a file-system source of truth to eliminate hallucinations and context bloat. It continuously scans dozens of sources including NVD, CISA KEV, arXiv and research blogs, then applies the custom TIPS scoring metric to rank threats before attempting reproduction. A dedicated Paper-to-PoC skill converts academic attack descriptions into executable code that is validated inside an isolated Ollama and Docker sandbox. Eight mandatory validation gates must pass before any finding is accepted, ensuring reproducibility and freshness of data. The system already supports tools such as Garak, PyRIT, promptfoo, fickling, modelscan and presidio, and evolves autonomously by committing successful attack scenarios back to its repository.
Anthropic Releases Eight Claude Code Updates in August Focused on Multi-Agent Workflows
Between August 13 and 21, Anthropic shipped eight consecutive Claude Code releases from version 2.1.232 to 2.1.239. The updates center on enabling multiple long-running agents that can share context, communicate across sessions, and continue work automatically after hitting usage limits. Key additions include subagent forking that inherits prompt cache and conversation history, cross-session messaging via @mentions, and an automatic mode that uses a classifier model to approve actions. Additional improvements cover GitLab merge request integration, memory management fixes for extended sessions, and support for native add-ons in musl-based Alpine environments. The changes significantly expand the scale of tasks that can be delegated to Claude Code without constant human oversight.
From Japanese sources
Translated from Japanese
CISA Adds Zimbra Collaboration Suite CVE-2026-73570 to KEV Catalog After Confirmed Exploitation
US authorities have issued a warning about active exploitation of a vulnerability in the Zimbra Collaboration Suite. CISA added the OS command injection flaw CVE-2026-73570 to its Known Exploited Vulnerabilities catalog on August 21, 2026, marking the fifth such entry for the year. The issue resides in the SNMP monitoring component and allows unauthenticated attackers to execute arbitrary operating system commands with Zimbra user privileges when SNMP notifications are enabled. MITRE rated the vulnerability 8.9 on CVSS v3.1, classifying it as High severity. Federal agencies must apply mitigations by August 24, while all organizations using the product are urged to prioritize patching to prevent further abuse.
IBM Patches 191 Vulnerabilities in AIX and PowerVM VIOS Including Critical Remote Command Execution Flaws
IBM disclosed a large set of security vulnerabilities affecting its UNIX-based AIX operating system and the PowerVM VIOS virtualization component. The company addressed a total of 191 CVEs through multiple updated releases for both products. Thirty-five of the issues received CVSS base scores of 9.0 or higher, with three OS command injection vulnerabilities rated at 9.9 that allow remote attackers to execute commands with low privileges. IBM provided specific Technology Level and Service Pack updates for AIX 7.3 and 7.2 branches along with corresponding fixes for PowerVM VIOS versions 4.1.x. The advisory covers both IBM-developed code and third-party software components integrated into the platforms. All listed CVEs are now resolved in the recommended maintenance levels released on August 15.
PostgreSQL Releases Security Update Fixing 28 Vulnerabilities and Over 110 Bugs
The PostgreSQL development team issued a major security update on August 13, 2026, addressing 28 vulnerabilities along with more than 110 bugs. While no issues reached a CVSSv3.1 base score of 9.0 or higher, 18 vulnerabilities scored 7.0 or above, with 14 rated at 8.8. Several flaws enable arbitrary code execution, including heap buffer overflows in regular expression processing and the to_char function. Additional fixes cover heap buffer overflows in pg_stat_statements and pg_dump, type confusion issues, and an SQL injection vulnerability. The update resolves specific CVEs such as CVE-2026-14664, CVE-2026-14669, CVE-2026-14670, CVE-2026-16238, CVE-2026-16239, and CVE-2026-15741.
Vulnerability in Docker go-archive Library Allows File Creation and Overwrite Outside Target Directory
A vulnerability has been identified in the go-archive archive processing library used by Docker and related software. The flaw, tracked as CVE-2026-17106, affects the Unpack, UnpackLayer, and Untar functions and permits files to be created or overwritten outside the intended extraction directory. The issue stems from a mismatch between the path string validated by the code and the actual path resolved by the operating system. Docker rated the vulnerability 7.1 under CVSS v4.0 and classified it as High severity. A proof-of-concept exploit has already been published for macOS and Linux. The fix was implemented in go-archive 0.3.0 and shipped in Docker Engine 29.7.0, Docker CLI 29.7.0, and Docker Desktop 4.86.0.
CISA Adds Two Remotely Exploitable TrueConf Server Vulnerabilities to KEV Catalog
The US Cybersecurity and Infrastructure Security Agency has added two critical vulnerabilities affecting TrueConf Server to its Known Exploited Vulnerabilities catalog. CVE-2026-72529 allows unauthenticated remote attackers to execute arbitrary scripts due to missing authentication in a critical function. CVE-2026-72530 is a code injection flaw that enables attackers to run arbitrary code on the underlying host system by escaping the sandboxed environment. Both issues can be exploited over TCP port 4307 without requiring authentication. Kaspersky assigned CVSS v3.1 base scores of 9.8 and 9.0 respectively, rating both as Critical. US federal agencies must apply mitigations for the first vulnerability by August 23 and for the second by September 3.
Oracle Releases August 2026 Monthly Security Patches Fixing 943 Vulnerabilities
Oracle has published its monthly Critical Security Patch Update on August 18, 2026, addressing a total of 943 vulnerabilities across a wide range of products. This release supplements the company's quarterly Critical Patch Update and includes fixes for third-party software issues, resulting in 925 unique CVEs after removing duplicates. Of these, 710 vulnerabilities received CVSSv3 base scores of 7.0 or higher, with 154 scoring 9.0 or above, including three at the maximum 10.0. A total of 467 flaws can be exploited remotely without authentication. Major products affected include Oracle Fusion Middleware with 262 patches, Oracle E-Business Suite with 120 fixes, and Oracle Database Server with six updates. The next monthly update is scheduled for September 15, 2026, followed by the quarterly release on October 20.
From Portuguese sources
Translated from Portuguese
LockBit Claims Breach of U.S. Bank, Sets September 2026 Deadline for Ransom Payment
The ransomware group LockBit has added U.S. Bank to its leak site, claiming to have stolen data from the financial institution and threatening to publish it unless a ransom is paid by September 3, 2026. U.S. Bank stated it is actively investigating the claims but has so far found no evidence of unauthorized network access or impact on internal systems. Lee Henderson, the bank's vice president of public affairs, confirmed that the institution is monitoring the situation and treating the allegations seriously. LockBit placed the bank's domain on its extortion portal on August 19 and has a history of resuming operations after a major international law enforcement disruption in 2024. The investigation remains ongoing, with the final scope depending on whether unauthorized access and data theft can be confirmed.
Chinese Cybercriminals Deploy AI Agents to Automate Exploitation of Exposed Windows and Linux Servers
A Chinese cybercriminal group tracked as UAT-10147 is actively using AI agents to automate reconnaissance, exploitation, and post-exploitation against internet-facing Windows and Linux servers. Cisco Talos identified the campaign after discovering approximately 170,000 URLs in the attackers' infrastructure, with targets spanning government, education, media, technology, and gaming organizations across multiple countries including Brazil. The group divides targets into smaller batches to accelerate scanning and exploitation of known vulnerabilities in applications such as Zimbra, AjaxPro, Nacos, and Telerik. AI agents go beyond simple code generation by producing exploitation guides, refining failed attacks, verifying payload success, and automating persistence and post-exploitation steps. In one observed case, the agents created scripts capable of validating permissions, exploiting ViewState deserialization, installing implants, and deploying web shells on compromised IIS servers. Cisco Talos assesses the operation as a shift toward semi-autonomous offensive orchestration rather than basic AI-assisted scripting.
AWS Details Architecture to Reduce Prompt Injection Risks in AI Agents
AWS has introduced a new architecture designed to prevent compromised or manipulated AI agents from accessing data beyond user permissions. The approach relies on Amazon Bedrock AgentCore to shift authorization decisions from the agent itself to the underlying infrastructure and connected services. The core risk arises when agents receive broad credentials to query databases, repositories, and SaaS platforms, allowing potential prompt injection attacks to retrieve unauthorized information. In the proposed design, users authenticate via Amazon Cognito and receive JWT tokens containing attributes such as department or role. The AgentCore Runtime validates these tokens before executing any agent actions, rejecting requests that violate configured rules. For DynamoDB queries, temporary credentials are issued through AssumeRoleWithWebIdentity, with IAM policies enforcing strict access to authorized data partitions only.
Manic Android Malware Steals PINs via Transparent Overlay and Relays Data Through Nearby Infected Devices
A newly identified Android malware strain named Manic merges banking trojan, spyware, and remote access capabilities. The threat has been active since at least February 2026 and continues to receive updates that add anti-analysis features, in-memory code loading, and lock-screen credential theft. Manic monitors 169 financial, messaging, and government applications while using a transparent overlay on legitimate numeric keyboards to capture PINs without displaying a full fake banking screen. Stolen data can be forwarded through other compromised nearby devices even when the original phone lacks internet connectivity. Operators also leverage WebRTC sessions for live screen viewing and remote interaction. The malware additionally functions as a keylogger, intercepts SMS and notifications, and collects passwords, one-time codes, and recovery phrases. Security researchers recommend avoiding unknown APKs and scrutinizing requests for Accessibility Services or broad device control permissions.
Oracle Issues Emergency Patches for Critical Remotely Exploitable WebLogic Server Vulnerabilities
Oracle released an emergency security update on August 18 containing 943 new patches across dozens of enterprise product families. The update addresses multiple high-severity flaws in Oracle WebLogic Server that can be exploited remotely without authentication over IIOP, T3, and RMI protocols. Four vulnerabilities—CVE-2026-60698, CVE-2026-60672, CVE-2026-60696, and CVE-2026-60977—received CVSS scores of 9.8 and can impact confidentiality, integrity, and availability of affected servers. A separate critical issue, CVE-2026-61241, was disclosed in the LDAP server component of Oracle Internet Directory with a maximum CVSS score of 10.0 and affects versions 12.2.1.4.0 and 14.1.2.1.0. Oracle urges customers to maintain supported releases and apply patches promptly, noting prior incidents where unpatched systems were targeted after fixes became available. Organizations are advised to validate patches in test environments before production deployment to minimize operational risk.
Google Patches Two Critical Memory Corruption Flaws in Chrome WebGL and Dawn Components
Google has issued a security update for Chrome that addresses 15 vulnerabilities, two of which are rated critical. The flaws, tracked as CVE-2026-76034 and CVE-2026-76036, involve buffer overflow conditions that can lead to out-of-bounds memory writes. CVE-2026-76034 affects the WebGL component used for 2D and 3D graphics rendering on web pages, while CVE-2026-76036 impacts Dawn, the Chromium implementation of WebGPU. Both issues were discovered internally and could result in crashes or remote code execution in certain scenarios. Updated versions are now available for Windows, macOS, Linux, and Android, and users are strongly advised to apply the patches immediately.
From Chinese sources
Translated from Chinese
NVIDIA Accelerates Physical AI Push with $500 Billion Infrastructure Plan at 2026 World Robot Conference
The 2026 World Robot Conference opened with 373 companies showcasing over 3,000 exhibits and more than 300 new products focused on embodied intelligence. NVIDIA is deepening its commitment to physical AI by partnering with Apollo, Blackstone, KKR and other major asset managers to create an independent financing platform targeting over $500 billion in third-party capital for AI infrastructure. Madison Huang, NVIDIA’s Senior Director of Physical AI and Robotics, visited the event to review advances in human data, simulation, and real-world deployment. The company is reframing compute resources as revenue-generating AI factories that produce tokens, simulation data, and action policies rather than treating them as cost centers. Humanoid robot development faces a critical bottleneck because high-fidelity physics simulation for millions of virtual agents demands enormous cloud compute, while sim-to-real transfer gaps and edge inference constraints remain unsolved engineering challenges.
Poisoned Rust Crates Execute Malware at Build Time: 245 Million Downloads Hit in Supply-Chain Attack
Three widely used Rust crates on crates.io were poisoned on August 20 with malicious versions that execute automatically during cargo build. The attack leveraged a typosquatted proc-macro1 dependency containing a build script that downloads payloads and establishes persistence. arrayref alone has accumulated 245 million downloads and is pulled automatically through caret ranges in many dependency trees. Attack infrastructure overlaps with prior campaigns attributed to Sapphire Sleet and MIDNIGHT NEPTUNE. Rust security teams yanked the malicious releases within 86-107 minutes, but the incident highlights missing publish-age controls and weak maintainer-account protections in the Cargo ecosystem.
Microsoft Defender Zero-Day ShieldBreak Enables SYSTEM Privilege Escalation, Bypassing July Rogue Planet Patch
Security researcher Nightmare Eclipse has publicly disclosed a new zero-day vulnerability named ShieldBreak in Microsoft Defender that allows any standard Windows account to escalate privileges directly to SYSTEM level. The flaw exploits a TOCTOU race condition in Defender's file scanning API callbacks, enabling attackers to swap file contents during the brief window when the scanner reads the file. Although linked to the earlier Rogue Planet vulnerability (CVE-2026-50656), the July patch did not fully address the attack surface, and an adjusted proof-of-concept continues to work on patched systems. Defender's high privileges and trusted status in enterprise environments amplify the risk, as compromised processes can execute malicious actions that appear legitimate in logs. Public disclosure without prior private reporting leaves a window of exposure until Microsoft issues an official fix. Blue teams are advised to enforce strict local permissions, monitor for anomalous Defender process behavior such as unusual child processes or repeated file handle operations, and implement layered defenses including application whitelisting and network segmentation.
Sorry Ransomware Exploits cPanel Vulnerability to Directly Lock Linux Servers in Multiple Chinese Incidents
China's National Computer Virus Emergency Response Center has issued a warning about the Sorry ransomware, which targets exposed Linux web servers through a cPanel authorization vulnerability. The Go-based malware gains root access without any phishing or user interaction, disguises itself as the legitimate sshd process, and follows a six-stage attack chain that includes data exfiltration before encryption. It terminates databases, security tools, and backup services, then uses AES and RSA to encrypt files with a .sorry extension while demanding ransom via an encrypted communication tool. The campaign specifically affects small and medium-sized enterprises running cPanel on mainstream Linux distributions, including domestic Xinchuang systems. Attackers also scan internal networks for weak SSH credentials to spread laterally. The center urges immediate patching of cPanel, exposure reduction, strong passwords, offline backups, and avoidance of fake decryptors.
Ruishu Information Warns Machine Traffic Now Dominates Internet as AI Agents Surge
Ruishu Information has released its 2026 Automation Threat Report covering data from early 2025 through Q2 2026. The report shows bots accounting for 68 percent of total internet traffic, with malicious bots making up 55 percent of that volume. Human traffic has fallen to just 22 percent while AI Agent-driven requests have grown from under 1 percent to 8-12 percent. LLM and AI Agent requests have already exceeded 450 billion, marking more than 400 percent year-over-year growth. The report introduces a new classification of non-human traffic into traditional bots, AI-enhanced bots, and autonomous AI Agents, along with an L1-L5 threat framework. It also expands documented attack scenarios from nine to thirteen, adding LLM application attacks, agent supply-chain attacks, identity hijacking, and autonomous AI-orchestrated attacks.
360 Launches nanoWork Enterprise Edition Channel Ecosystem with Native Security for AI Agents
On August 17, 360 held its nanoWork Enterprise Edition partner recruitment launch event, gathering over 200 channel partners from across China. The event marks the first major channel gathering since the product's official release on July 28. 360 founder Zhou Hongyi emphasized that AI is moving from answering questions to completing tasks, requiring robust security foundations, channel networks, and service systems. General Manager Wang Yi highlighted the low enterprise AI adoption rate of around 18% and positioned nanoWork as a secure, enterprise-grade AI agent platform built on 20 years of 360 security expertise. The company unveiled a three-dimensional partner architecture, four revenue models, and five empowerment systems to enable lightweight partner entry into the market. With a 1,000+ FDE engineer network and the AikerWorld community, nanoWork aims to deliver on-site陪跑 and service packages to bridge the last mile of AI deployment.
From Spanish sources
Translated from Spanish
CISA Adds Actively Exploited Zimbra CVE-2026-73570 to KEV Catalog with Immediate Patching Deadline
CISA has added CVE-2026-73570 affecting Zimbra Collaboration Suite to its Known Exploited Vulnerabilities catalog, ordering urgent remediation due to confirmed active exploitation. The flaw enables unauthenticated remote command execution on servers where the optional zimbra-snmp package is installed and SNMP notifications are enabled. Exploitation occurs via specially crafted SMTP requests that trigger operating system command injection under the privileges of the zimbra user. Zimbra released the fix in version 10.1.20 on July 20, 2026, while CISA placed the vulnerability in the KEV list on August 21 and set an August 24 deadline for federal civilian agencies. Organizations are advised to first verify exposure by checking for zimbra-snmp and SNMP notification settings before applying the update. Post-patch steps include validating the version, controlled service restarts, and hunting for indicators such as unexpected Zimbra restarts, anomalous zimbra user activity, and new files in sensitive directories like /opt/zimbra/jetty/webapps/. The directive highlights the critical nature of pre-authentication flaws in internet-facing mail servers that process untrusted inbound traffic.
Critical GeoServer Vulnerability CVE-2024-36401 Actively Exploited for Unauthenticated Remote Code Execution
The CVE-2024-36401 flaw in GeoServer and its GeoTools library allows attackers to achieve remote code execution without authentication by abusing property name expressions interpreted as XPath. Active exploitation has already led to confirmed intrusions involving initial access, lateral movement, and persistence with tools such as China Chopper web shells. Multiple OGC endpoints including WFS GetFeature, WMS GetMap, and WPS Execute are affected when exposed to the internet. Patches are available in GeoServer versions 2.22.6, 2.23.6, 2.24.4, and 2.25.2. Organizations unable to patch immediately can mitigate risk by removing the gt-complex JAR file, though this may break functionality. Additional defenses include restricting internet exposure through IP allowlisting, VPNs, and reverse proxies while monitoring logs for anomalous requests. Any previously exposed instances should be treated as potentially compromised, with full incident response including credential rotation and host forensics recommended.
Critical GeoServer Vulnerability CVE-2024-36401 Actively Exploited for Unauthenticated Remote Code Execution
The vulnerability CVE-2024-36401 in GeoServer is already being exploited in real-world attacks and enables remote code execution without authentication. The flaw stems from how GeoServer and its GeoTools library interpret certain property names, allowing malicious XPath expressions via commons-jxpath in default configurations. Attackers can abuse multiple OGC endpoints including WFS GetFeature and GetPropertyValue, WMS GetMap, GetFeatureInfo and GetLegendGraphic, plus WPS Execute to gain initial access. Observed intrusions follow a familiar pattern of reconnaissance, lateral movement and persistence with web shells such as China Chopper. Patches are available in GeoServer 2.22.6, 2.23.6, 2.24.4 and 2.25.2, while a temporary mitigation involves removing the gt-complex jar file. Organizations are urged to apply updates immediately, restrict internet exposure and hunt for indicators of compromise in logs and on hosts.
Attackers Actively Exploit Critical SharePoint Authentication Bypass CVE-2026-55040 After Public PoC Release
Active exploitation of CVE-2026-55040, a critical authentication bypass vulnerability in Microsoft SharePoint Server, has intensified following the public release of a proof-of-concept exploit. The flaw, which carries a CVSS score of 9.1, stems from multiple defects in the JWT token validation pipeline and allows remote attackers to impersonate any user, including administrators, without valid credentials. Affected versions include SharePoint Server Subscription Edition, SharePoint Server 2019, and SharePoint Server 2016, with heightened risk for instances exposed to the internet. Microsoft released patches in July 2026, and organizations are urged to apply them immediately while restricting external access and monitoring IIS logs for anomalous administrative activity. The vulnerability has already been chained with other flaws in real-world incidents involving persistence, data theft, and credential rotation responses.
Attackers Exploit Critical CVE-2026-59310 in VMware vCenter for Persistent Remote Access
A critical vulnerability identified as CVE-2026-59310 in Broadcom VMware vCenter Server is being actively exploited in the wild against internet-exposed instances. The flaw resides in the Syslog server component and enables remote code execution through a path traversal weakness, carrying a CVSS score of 9.8. Attackers have been observed deploying malicious cron jobs and the reverse_ssh tool to establish persistent outbound command-and-control channels since early August 2026. The campaign has impacted 361 unique IP addresses across 47 countries, with notable concentrations in Germany, the United States, Turkey, Iran, and France. Broadcom has released patches under advisory VMSA-2026-0006.1, which also addresses the related CVE-2026-59309, and strongly recommends immediate updates along with network segmentation and log reviews. No workarounds exist, making prompt patching the only effective mitigation.
GhostSplice Technique Lets Malicious MCP Servers Trick AI Coding Agents into Exfiltrating Secrets
GhostSplice is a new technique that allows a malicious MCP server to induce an AI coding agent to leak SSH keys, environment secrets, and source code. The attack splits malicious instructions across tool metadata and responses so the agent reconstructs and executes the full exfiltration plan without detecting an overtly malicious command. Tests showed the method raised compliance rates from an average of 42 percent to 82 percent across eleven models, with some systems moving from zero to 100 percent success. The technique requires the developer to connect the attacker-controlled MCP server and for the agent to already possess read access to the targeted files. Defenses focus on strict allow-listing of MCP servers, least-privilege tool permissions, separation of tool output from instructions, and human approval for sensitive operations. The disclosure aligns with prior warnings about poisoned MCP tool descriptions and agentjacking attacks.