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
AWG Router Delivers Script-Free AmneziaWG Cascades with Load Balancing, Speed Limits and Leak Prevention
AWG Router is an open-source Python-based web panel that manages complex AmneziaWG VPN cascades without hundreds of lines of bash scripts. The tool generates nftables rules, ip rules, and tc queues to route Russian networks directly while sending other traffic through foreign AmneziaWG servers or SOCKS5 proxies. It enforces strict leak protection so that traffic never falls back to the entry server’s Russian IP when an upstream tunnel fails. The panel supports per-client routing policies, domain-based rules via SNI inspection, automatic load balancing across multiple exits, and granular speed limits using HTB and fq_codel. All state is stored in resilient JSON files, and the system continues operating even if the panel itself is restarted or updated. The project is released under the MIT license with full source available.
Translated from Russian
Nightmare Eclipse Publishes SignDrop Zero-Day in Windows CLFS Driver
Security researcher Abdelhamid Naseri, known as Nightmare Eclipse, has disclosed a new zero-day vulnerability named SignDrop affecting the Common Log File System (CLFS) driver in Windows. The flaw allows attackers to bypass authentication protections introduced by Microsoft in October 2025 and escalate privileges on affected systems. SignDrop exploits the way CLFS handles .cnpf authentication file creation when a new log file is generated, enabling a privileged process to open the file later and create it with SYSTEM rights. The technique can also be abused against any antivirus or EDR engine to disable self-protection mechanisms by making file creation requests appear to originate from the legitimate security process itself. Nightmare Eclipse confirmed the issue works on Windows 11 26H2, Windows 11 26H2 Experimental, and Windows Server 2025 Experimental, impacting an estimated 80 percent of Windows installations. No workaround exists other than waiting for a Microsoft patch, as disabling CLFS protections would reintroduce memory corruption risks. The researcher previously released multiple other zero-days, including BigDiskBuster and RoguePlanet targeting Microsoft Defender.
Translated from Russian
Anthropic Funds New Security Developer Role at Python Software Foundation Following Sharp Rise in Vulnerabilities
Anthropic has financed a new security position within the Python Software Foundation in response to a marked increase in vulnerabilities and malicious packages affecting the Python ecosystem. Data published by Seth Larson showed a significant spike in reported issues during 2026. Brett Smith has been appointed as Security Developer and will collaborate with Larson, who serves as Security Developer-in-Residence, and Mike Fiedler, the PyPI security engineer. The expanded team will focus on scaling automated quarantine mechanisms in PyPI and streamlining the workflows of the Python Security Response Team. At PyCon US 2026, the team reported handling three to four times more vulnerability disclosures for Python and pip compared with the previous year, alongside a 60 percent increase in malware detections and removals from the repository.
Translated from Russian
Detecting Deepfakes: Tools, Signs, and Verification Methods for Video, Photos, and Voice
Deepfake technology now creates highly convincing synthetic videos, photos, and audio that can impersonate real people in video calls or voice messages. The article explains how modern neural networks using GANs, diffusion models, and transformers generate face swaps, voice clones, and fully synthetic scenes. It details visual and audio artifacts that may reveal fakes, such as inconsistent lighting, deformed fingers, or unnatural speech pauses, while noting that these signs are becoming less reliable. Practical verification steps include reverse image searches, metadata analysis, and tools like InVID-WeVerify. Specialized detectors such as DeepFake-o-Meter, Reality Defender, and Sensity AI are reviewed along with their limitations against unknown generators. Real-world cases, including a $25 million corporate fraud using deepfake video calls, illustrate the growing risks and the need for independent confirmation rather than relying solely on automated tools.
Translated from Russian
Uber Confirms Data Breach Exposing Drivers' License Records via Contractor Fireworks AI
Uber has notified thousands of drivers that their personal information and driver's license details were stolen during an October incident at one of its technology contractors. The breach occurred at Fireworks AI, a startup providing access to open-source AI models, after attackers compromised internal credentials. Stolen data includes driver's license information affecting 2,567 residents in Massachusetts, 3,430 in South Carolina, and 1,873 in Rhode Island, with additional notifications sent in Illinois and California. Fireworks AI also confirmed the compromise of client data, including payment details and private AI training datasets, impacting other customers such as Notion. The attackers are believed to be financially motivated and were targeting tokens for large language models. This marks Uber's second reported data incident this year following the earlier Uber Freight breach.
Translated from Russian
Roskomnadzor Deploys 2.27 Billion Ruble AI Traffic Filter: Known Facts and Expert Speculations
Russian regulator Roskomnadzor has integrated a machine learning-based traffic filtering system into the automated security infrastructure of the Russian internet segment. The system, valued at 2.27 billion rubles, operates on domestic models within the agency's own infrastructure and connects to TSPU equipment that performs DPI-based filtering. Publications including Forbes, Meduza, 3DNews and SecurityLab base their reporting on the 2026 digital transformation passport, which also allocates additional billions for TSPU upgrades and hardware. Experts suggest the model analyzes packet sizes, session durations, traffic ratios and server characteristics rather than relying solely on known signatures, though no official performance metrics have been released. The exact role of the model—real-time analysis versus signature generation—remains unconfirmed, as does its impact on VPN protocols and potential for false positives affecting corporate and cloud traffic.
Translated from Russian
Can Open-Source AI Agents Replace Pentesters? Strix and PentAGI Tested on Real 200-Host Perimeter
An independent researcher evaluated two open-source LLM-based pentesting agents, Strix and PentAGI, against a live external perimeter containing roughly 200 active hosts. Both tools autonomously planned tasks, executed CLI utilities, generated proofs of concept, and produced reports while running inside isolated Docker sandboxes. Strix delivered only 15 findings with almost no false positives yet consistently overstated severity, while PentAGI reported 296 vulnerabilities including 42 critical issues, none of which proved exploitable. Real critical vulnerabilities on the same scope, including an RCE chain, were discovered only by a commercial autonomous scanner. The experiment highlighted fundamental limitations in scope control, context management, and risk assessment that still require human oversight. Token consumption reached 145.1 million for Strix and 31.4 million for PentAGI, illustrating the cost of maintaining long-running autonomous operations.
Translated from Russian
Beget Hosting Provider Restores 95% of Virtual Hosting Servers After Destructive Cyberattack
Russian hosting provider Beget has recovered a significant portion of its infrastructure following a cyberattack that occurred in the night between November 1 and 2, 2026. Attackers gained access to the management plane and executed wiping scripts, resulting in complete data loss across the cloud infrastructure including VPS, S3, and Kubernetes environments. File-level backups of VPS instances were preserved and are now being used for recovery. The company has restored 95% of virtual hosting servers and 94% of dedicated servers while gradually re-enabling domain management functions and the control panel. Beget declared force majeure and temporarily lost accreditation as a registrar for .RU and .RF domains. Technical support has been partially restored through the control panel for affected customers, with responses taking longer due to high volume of requests.
Translated from Russian
Malicious ZIP Archive Delivers Git-Based Remote Code Execution via Hooks and Filters
A developer received a fake job offer from Cryptic Heaven through LinkedIn, leading to a ZIP file containing a rigged Git repository. The archive included a pre-configured .git directory with altered hooksPath pointing to .git/objects/interrupt and a custom clean filter named ltr. Upon git status or branch checkout, the setup triggered a shell script that downloaded and executed remote code from a server hosted on vercelscript.vercel.app. The attack leveraged post-checkout hooks and Git filters to run silently without requiring install scripts or explicit execution. Multiple trigger paths existed, including automatic detection in Visual Studio Code, Git-aware terminal prompts, and following README instructions to switch branches. The technique worked on both Linux/macOS and Windows through Git for Windows shell support. The incident highlights risks of opening untrusted Git repositories delivered outside standard clone workflows.
Translated from Russian
From Russian sources
Translated from Russian
How a Personal Site Counts Bots Without Storing Any IPs: Pure PHP, No Database, Strict CSP and 160 Tests
A personal website built entirely in pure PHP without frameworks or databases demonstrates advanced defensive security practices while tracking bot activity from an SSH honeypot. The site stores all content in a single JSON file with atomic writes and automatic backups, eliminating database-related attack surfaces and simplifying deployment to folder copying. Strict Content-Security-Policy rules prohibit inline styles and scripts without nonces, forcing the use of SVG attributes, CSSOM updates, and JSON data blocks instead of direct attribute injection. Over 160 automated tests written in PHP attack the site with XSS payloads, CSRF attempts, path traversals, and malformed Markdown to verify that no malicious output reaches the browser. Bot identification relies on behavioral signatures derived from the first twelve cleaned commands of each session, combined with HASSH fingerprints and activity patterns, allowing the site to report unique bot families and address counts without ever publishing IPs. Additional browser-based tests using Chromium validate accessibility, performance budgets, and CSP compliance across desktop and mobile viewports. The entire system runs with zero external dependencies and integrates tests directly into the deployment pipeline.
Telegram Patches RCE Vulnerabilities in HTML Widgets After Researcher Demonstrates Calculator Exploit
Telegram introduced HTML widgets for bots in early November, allowing JavaScript execution inside messages via the QuickJS engine. Security researcher V12 quickly demonstrated remote code execution by launching the system calculator on Android. Pavel Durov responded that no vulnerability existed because the user had to first start the bot, yet server-side fixes were deployed anyway. Multiple use-after-free and use-after-return bugs were identified in QuickJS and QuickJS-ng, some discovered with AI assistance from Claude Fable 5.1 and Mythos 5.2. Telegram isolated the widget engine into a separate process on Android, added syscall filtering, and limited nested graphical effects to prevent crashes. Desktop clients received the fixes across versions 7.4.2 and 7.4.3, while widgets were temporarily restricted server-side. Users are advised to update to Android 13.2.2 or Desktop 7.4.3 and avoid unknown bots.
Telegram Patches Critical QuickJS Sandbox Escape Vulnerability in Android Update 13.2.2
Telegram has released version 13.2.2 for Android to address a high-severity vulnerability in its HTML widgets for bots. The flaw resided in the QuickJS JavaScript engine within Telegram's QuickJS-ng fork and allowed scripts in ordinary messages to break out of the sandbox and execute native code on the device. The root cause was improper handling of the await using construct inside non-async functions, which triggered a use-after-return condition by freeing local variable references while nested functions still accessed the stack. The vulnerable code entered the QuickJS-ng main branch on May 16 as part of explicit resource management support. Security researcher Andreas Rösler discovered the issue during an AI-assisted audit with Anthropic's Claude model that identified 27 findings total. The patch enforces a syntax error on the invalid construct and was merged on November 10, the same day the Android release became available. Although reported against Android 13.2.1, the bug affected every platform using the affected engine fork.
Russian Finance Ministry Seeks Central Bank Reports on Systematic P2P Transfers to Tax Authorities
The Russian Ministry of Finance has proposed amendments requiring the Central Bank to share data with the Federal Tax Service on individuals whose bank accounts show signs of regular income received from other private citizens. The changes target article 85 of the Tax Code and form part of the budget package bill numbered 617-9. Under the new rules, the Central Bank would identify such citizens through transaction analysis and transmit the information according to a future agreement with the tax service. The Federal Tax Service would also gain expanded powers to request bank statements from non-entrepreneurs without needing prior debt collection or audit triggers. If approved, the provisions would take effect on 1 January 2027 and apply to operations conducted after 1 January 2026. Separate bills under discussion would impose monthly transfer limits of 30,000 rubles on suspected clients starting in 2028 and require the tax service to notify the Central Bank about account openings and income categories.
Negotiating with Ransomware Attackers: Lifeline or Legal Trap?
Ransomware incidents leave organizations facing encrypted systems and threats to leak stolen data, prompting urgent questions about whether to contact extortionists. Experts emphasize that initial response must focus on containment, evidence preservation, and backup verification rather than immediate payment discussions. NCSC guidelines stress isolating affected systems, securing communication channels, and understanding the exact nature of demands before any dialogue begins. Real-world cases such as the PowerSchool breach demonstrate that paying ransoms does not ensure data deletion or end extortion attempts. Legal risks vary by jurisdiction, with prohibitions in places like North Carolina and sanctions concerns from OFAC complicating decisions for many entities. Technical recovery options, including tools from the No More Ransom project, should be evaluated alongside insurance and regulatory notification requirements under Russian Federal Law No. 152-FZ. Ultimately, successful resolution depends on verified restoration capabilities and eliminating attacker persistence rather than any negotiated discount.
MinTsifry Proposes Requiring Telecom Operators to Install SORM Equipment Before Obtaining Licenses
The Russian Ministry of Digital Development plans to introduce legislation obligating telecom operators to deploy SORM surveillance systems prior to receiving operating licenses. Under the proposed rules, companies will only be permitted to begin serving subscribers after the equipment has been installed and handed over to law enforcement agencies. Non-compliance could result in license revocation through an extrajudicial administrative process. Current regulations allow operators to obtain licenses with a network diagram that merely indicates planned or existing SORM components. The change aims to close loopholes exploited by unscrupulous providers that repeatedly register new legal entities to operate without surveillance capabilities. Industry experts note that the minimum cost of a basic SORM complex starts at 5 million rubles, which could disproportionately affect smaller local providers without a transition period or scaled requirements.
From Japanese sources
Translated from Japanese
Fortinet Discloses Multiple Critical Vulnerabilities in FortiMail, All Addressed in October Update
Fortinet has published 24 security advisories detailing 27 vulnerabilities affecting its FortiMail email security product. Eight of the flaws received the highest severity rating of Critical, with three carrying a CVSSv3.1 base score of 9.8. The affected issues include path traversal CVE-2026-126147, command injection CVE-2026-126148, and SQL injection CVE-2026-126263, each capable of enabling arbitrary code execution. An additional unauthenticated SQL injection vulnerability tracked as CVE-2026-126164 scored 9.4, while four more flaws were rated 9.1. All issues were resolved in the October security update released by the vendor prior to the November 10 disclosure. No exploitation of the vulnerabilities has been observed in the wild at the time of advisory publication.
Cloud Software Group Patches 28 Vulnerabilities in Citrix NetScaler ADC and Gateway
Cloud Software Group has released security updates for Citrix NetScaler ADC and Citrix NetScaler Gateway to address multiple critical vulnerabilities. The company also disclosed a critical flaw in Citrix NetScaler Console. A total of 28 vulnerabilities were identified across the products, with several receiving CVSS v4.0 base scores as high as 9.5. The most severe issue, CVE-2026-107412, involves memory overflows affecting gateway and AAA virtual servers. Eight additional flaws, including CVE-2026-120868, scored 9.3 and impact standard packet processing. Other affected areas include ICA proxy handling, VPN endpoints, mail OTP authentication, and DNS components.
Ivanti Patches 13 Vulnerabilities in Five Products, Including Two Critical Flaws
Ivanti has issued security updates for five enterprise products to address a total of 13 vulnerabilities, two of which carry a critical severity rating. The affected products include Ivanti Endpoint Manager, Ivanti Cloud Services Application, Ivanti Sentry, Ivanti Endpoint Manager Mobile, and Ivanti User Workspace Manager. The most severe issues allow unauthenticated remote code execution and unauthorized certificate manipulation. One critical flaw, CVE-2026-121653, affects the Core component of Ivanti EPM and scores 9.8 on CVSS v3.0. A second critical vulnerability, CVE-2026-121683, impacts Ivanti CSA version 5.0.8 and enables arbitrary addition or deletion of certificate registration data. The remaining vulnerabilities are rated high and affect multiple other Ivanti solutions.
Google Issues Second Chrome Update This Week, Patching 83 Vulnerabilities Including Five Critical Flaws
Google has released a new security update for its Chrome browser, marking the second release within the same week. The update addresses a total of 83 vulnerabilities across Windows, macOS, and Linux platforms. Five of these issues are rated Critical and involve Use After Free memory errors in key components such as ServiceWorker, CSS, Layout, Skia, and Accessibility. The previous update from November 9 fixed nine vulnerabilities, one of which (CVE-2026-125350) had already been observed in active exploitation. No evidence of in-the-wild exploitation has been reported for the latest set of fixes. The new versions are Chrome 157.0.8103.79 for Windows and macOS and 157.0.8103.78 for all three operating systems.
Mozilla Releases Firefox 160 Fixing 99 Vulnerabilities Including Multiple High-Severity Issues
Mozilla Foundation has published Firefox 160 along with its extended support release, addressing a total of 99 vulnerabilities. Although no Critical-rated flaws were present, 57 vulnerabilities received a High severity rating. Among the notable issues is a Use-After-Free flaw in CanvasWebGL (CVE-2026-124907) that could enable privilege escalation and sandbox bypass. A separate Use-After-Free vulnerability tied to inter-process communication (CVE-2026-124903) also carries the risk of sandbox escape. Additional patches resolve Same-Origin Policy bypasses tracked as CVE-2026-124874 and CVE-2026-124902, along with numerous other security problems across the browser.
Fortinet Discloses Eight FortiOS Vulnerabilities Including Three Critical Flaws
Fortinet has published security advisories detailing eight vulnerabilities affecting its FortiOS operating system used in UTM appliances. Three of the issues received the highest severity rating of Critical. CVE-2026-126259 allows unauthenticated command execution with super-admin privileges in FGSP combined with configuration synchronization or auto-scale setups via crafted HA packets. CVE-2026-126254 stems from integer truncation in the BGP daemon and can propagate malicious attributes from configured iBGP peers. CVE-2026-126253 enables authentication bypass in the OSPF daemon, permitting unauthenticated LSA injection. The remaining five vulnerabilities were assigned lower severity ratings. Administrators are urged to apply the available patches promptly.
From Portuguese sources
Translated from Portuguese
Qilin Ransomware Claims at Least 365 Victims in Three Months of Attacks
The Qilin ransomware group has claimed responsibility for at least 365 organizations compromised over a three-month period, according to research published on November 11, 2026. Operating as a ransomware-as-a-service platform since September 2022, Qilin supplies the attack infrastructure while affiliates conduct intrusions and receive at least 80 percent of ransom payments. The group employs double extortion tactics, stealing sensitive data before encryption and threatening public leaks if ransoms are not paid. Its affiliate program was described as the most active worldwide during the measured timeframe. Researchers also linked the group's administrator, known as Haise, to Aleksey Mikhailovich Yakovtsev, a 36-year-old resident of Saint Petersburg, Russia, through reused online identities spanning multiple cybercrime forums and leaked records from Gosuslugi. The investigation highlighted Qilin's technical reach, including targeted attacks against VMware vCenter and ESXi environments using remote management tools, lateral movement, and vulnerable drivers to disable security products.
Node.js Patches 16 Vulnerabilities Exposing TLS Private Keys and Cross-Connection Data
The Node.js project has released fixes addressing 16 vulnerabilities, seven of which are rated high severity. The most critical issue, CVE-2026-123864, allows an unauthenticated client to read memory from other HTTPS connections, potentially exposing TLS private keys and requests belonging to other users. Three supported release lines were affected, and exploitation requires a specific application code pattern involving res.flushHeaders() and res.end() calls before the request body is fully received. Additional high-severity flaws include CVE-2026-123863, which crashes HTTP/2 servers, CVE-2026-123866 that leaks server memory chunks to clients, and CVE-2026-123865 that causes permanent memory and file descriptor leaks in HTTP/2 servers. Several other CVEs affect TLS renegotiation, SNI callbacks, and file streaming functions. The project states that end-of-life branches including 20.x and 25.x remain vulnerable with no patches available.
Microsoft Patches Two BitLocker Vulnerabilities Allowing Encryption Bypass via Physical Access
Microsoft released fixes on November 10, 2026, for two BitLocker vulnerabilities that enable attackers with physical access to bypass disk encryption on Windows systems. CVE-2026-119832 carries a CVSS score of 5.9 and stems from an improperly applied access control, allowing low-privileged local users to read protected data from the system drive without user interaction. CVE-2026-97015 scores 6.8 and involves a buffer overflow in dynamically allocated memory, requiring no prior privileges and classified as low attack complexity. Both flaws affect Windows 10 versions 1809, 21H2 and 22H2, multiple Windows 11 releases, and Windows Server 2019, 2022 and 2025 editions across various processor platforms. The issues directly undermine BitLocker's core purpose of protecting lost, stolen or improperly decommissioned devices. Corresponding updates include KB5131096 for Windows 10 21H2/22H2 and KB5131087 for Windows 11 23H2, with administrators advised to verify coverage for CVE-2026-97015 separately.
Qlik Sense Servers Targeted with Reverse Shell Attacks in October 2026 Campaign
In October 2026, multiple Qlik Sense servers hosted by European Union customers were subjected to attempted reverse shell installations within a three-minute window. The attacks were blocked before execution and originated from a single-line PowerShell command delivered through the Qlik Sense scheduling component. Analysts noted strong similarities to a November 2023 campaign attributed to CACTUS ransomware operators that exploited three specific vulnerabilities: CVE-2023-41265, CVE-2023-41266, and CVE-2023-48365. The observed activity included a Go-based second-stage payload that searched for Apache ActiveMQ processes, harvested local account hashes on Linux, and deployed modified OpenSSH servers for persistence. Infrastructure recovered during the investigation also contained exploits for a 2023 Apache ActiveMQ vulnerability and a 2024 Veeam Backup and Replication flaw, indicating broader targeting intentions. The root vulnerability exploited in the 2026 attempts remains unidentified.
Notion Confirms Exposure of AI Assistant Transcripts After Fireworks AI Breach
Notion has confirmed that a security incident at Fireworks AI exposed conversation transcripts from its Notion AI assistant, affecting less than 1% of customer workspaces. The breach occurred because Fireworks handled classification tasks for the Insights feature and served open-weight models. Notion stated that its own systems were not compromised. The timeline shows initial notification on November 2, 2026, followed by disconnection of systems on November 3 and confirmation of customer data exposure on November 8. Most transcripts lacked author or workspace attribution and did not include tool call results, though user-entered content could still identify individuals or organizations. Fireworks attributed the intrusion to a financially motivated group engaged in LLMjacking, involving theft of credentials to hijack AI models for illicit use.
Ransomware Group BYOD Claims Theft of Toyota Customer Personal Data in Unverified Extortion Listing
A ransomware group calling itself BYOD has added Toyota to its victim list on a dark web extortion site, claiming to have stolen personal information belonging to the automaker’s customers. The announcement, updated on November 10, 2026, alleges access to full names, email addresses, physical addresses, telephone numbers, and dates of birth, along with data from the company’s self-service kiosk terminals. No independent verification of the claims has been performed, and the listing provides no details on the volume of records, affected regions, or specific business units involved. The group states that parts of the environment were isolated but claims residual remote access paths remained; however, the announcement contains no technical indicators or proof-of-compromise samples. Toyota has not issued any public statement confirming or denying the incident. The report notes that similar unverified listings are sometimes used to pressure victims into negotiations and may include repackaged data from prior breaches. This marks a contrast with Toyota’s 2023 public disclosure of a separate cloud misconfiguration incident that the company itself reported.
From Chinese sources
Translated from Chinese
Thousands of Grok-Generated Apps Expose Sensitive Corporate Data Through Default Public Access
Security firm Claroc scanned 43,551 Grok-built application addresses and found 37,571 still online, with 94.7 percent requiring no login. Approximately 650 to 800 of these apps contained sensitive internal data belonging to SpaceX, X, Tesla, and 30 other organizations, including six Fortune 500 companies. Exposed records included ICU patient reports with lab results and ventilator status, contractor timesheets containing plaintext passwords, and Tesla service-center performance evaluations listing employee names, efficiency metrics, and bonuses. Researchers demonstrated that Grok sometimes inverts permission-check logic, allowing unauthorized Google accounts to view and modify data after login. In contrast, ChatGPT-generated apps default to private settings, preventing access even when URLs are discovered via passive DNS. The incident highlights how AI coding tools lower the barrier to publishing internal applications while weakening the review processes that traditionally protected sensitive data.
Baidu Launches Ciyuan Exchange to Establish Trust Standards for AI Model Trading
Baidu Intelligent Cloud has introduced the Ciyuan Exchange, a dedicated trading platform for large AI models that aims to solve quality verification and credit issues in an increasingly crowded market. The platform, unveiled at the Baidu World 2026 Conference, already hosts over 100 supply resources with daily calls exceeding 50 million and daily token usage surpassing 700 billion. Founder Li Yanhong positioned the exchange as an infrastructure layer that moves beyond simple model aggregation toward verifiable identity, ongoing performance monitoring, and supplier credit records. The initiative addresses real procurement risks where identical models from different providers show up to twofold differences in throughput and reliability, according to a China Academy of Information and Communications Technology report. By combining model fingerprinting, capability testing, periodic audits, and continuous credit scoring, the platform seeks to create transparent market mechanisms similar to commodity exchanges. This development signals a broader industry shift from raw model capability competition toward standardized service quality, pricing transparency, and reliable delivery infrastructure.
Over a Dozen AI Platforms Breached in One Month for API Keys in LLM-Jacking Operations
Between late September and early November, at least a dozen AI developer platforms were compromised in a coordinated campaign targeting API access tokens rather than customer data. Affected organizations include Andromeda, NVIDIA Brev, CommandCode, Fireworks AI, Lightning AI, Northflank, OpenCode, Railway, Trigger.dev, Warp, Zeabur, Chainstack, and WorkOS. Attackers exploited third-party components and internal sales applications to reach stored credentials used for AI inference services. The stolen tokens enable LLM-jacking and model distillation activities, with multiple threat groups observed reusing the same credentials shortly after disclosure. The open-source penetration testing framework Artex was used to automate reconnaissance and exploitation, lowering the barrier for less skilled actors. Companies such as Chainstack and WorkOS confirmed the incidents through their own disclosures, highlighting risks in non-core systems that store high-value AI credentials.
Stolen Credentials Poison cPGuard Update Channel, Turning Security Software into Data Theft Backdoor
Attackers used compromised OPSSHIELD credentials to hijack the cPGuard update distribution network on November 2 at 02:06 UTC, injecting malicious scripts into versions 5.88.12 through 5.88.19. The poisoned updates targeted Linux servers running the security tool, which holds root privileges on hosting provider and cloud infrastructure machines. The script harvested /etc/shadow, /etc/passwd, MariaDB/MySQL root credentials from my.cnf, root shell history containing API tokens, and SSH authorized_keys files before exfiltrating them via the vendor's own infrastructure. A persistent backdoor was also dropped at /var/tmp/.cpg91511_sh. OPSSHIELD detected the intrusion within 12 hours and halted distribution, but confirmed that 32 customer servers had their data accessed. The incident highlights risks of relying on static credentials for security product update pipelines without code signing or behavioral restrictions.
Lost Control Series Highlights Realistic Social Engineering Attacks Disguised as Workplace Documents
A new micro-film in the Lost Control series demonstrates how employees can be tricked into compromising their organizations by opening files that appear to be routine work documents such as layoff lists, internal notices, or tax inspection alerts. The six-minute video recreates authentic attack vectors including phishing emails, fake download sites, bundled legitimate software packages, and phishing websites, as documented in five recent cases released by China's Ministry of Public Security Cyber Security Bureau. These attacks succeed because they mimic normal business communications rather than obvious malware delivery methods. The producers recommend sharing the film internally, reviewing work group messages from the past 30 days for suspicious attachments labeled as personnel lists or official notices, and verifying any such files before opening them. Organizations whose annual security training lacks measurable improvement in employee detection rates are invited to contact Allsec for assessment. Victims who have already clicked links or transferred funds are advised to call the national anti-fraud hotline 96110 immediately. The content is AI-generated with artistic modifications based on publicly reported incidents.
AI Agents Rewrite SaaS Monetization Models as Marketplace Platforms Drive Global ISV Growth
The 2026 narrative of SaaS apocalypse driven by AI has reversed following strong earnings from major vendors. Companies including Salesforce, Palantir, Kingdee and Yonyou began reporting AI-related revenue, while IDC and Nvidia CEO Jensen Huang argued that AI agents amplify rather than replace software. At the 2026 Amazon Web Services Software Enterprise Summit in Shenzhen, AWS launched an ISV go-global accelerator and six new AI agents on Marketplace to streamline sales, compliance and partner management. Marketplace now hosts over 30,000 products from 6,000 sellers, with Agentic AI listings growing from 800 to more than 3,000 since July 2025. Chinese ISVs such as PingCAP, Zilliz, Manus and Plaud AI are shifting from simple listings to localized operations and bundled solutions. The platform projects hundreds of billions in transaction volume for 2026, supported by a network of 130,000 global partners. Industry leaders emphasize that future growth depends on embedding AI as core product logic rather than add-on chat interfaces.
From Spanish sources
Translated from Spanish
Node.js Issues Security Releases for 26.x, 24.x and 22.x Fixing High-Severity TLS and Brotli Flaws
Node.js published security updates on October 21, 2026 for the 26.x, 24.x and 22.x release lines. The patches address multiple high-severity vulnerabilities affecting TLS hostname validation and the Brotli decoder, along with updates to critical dependencies including OpenSSL and Undici. Several CVEs cover bypasses in certificate hostname checks when IP addresses use full-width digits or when hosts resolve loosely to IP addresses. A particularly serious out-of-bounds write in the Brotli decoder (CVE-2026-111497) could enable remote code execution on services processing compressed responses. Additional fixes resolve keep-alive socket timing issues, DNS resolution truncation with NUL characters, and entropy caching problems in crypto.randomUUID and randomInt functions. The releases also update c-ares, Undici, OpenSSL, SQLite and ca-cert bundles following the eMudhra intermediate CA compromise.
Node.js Fixes High-Severity Brotli Decoder Flaw Leading to Remote Code Execution
A high-severity vulnerability in the bundled Brotli decoder of Node.js allows out-of-bounds writes when processing attacker-controlled compressed data, potentially resulting in remote code execution. The flaw, tracked as CVE-2026-111497, affects all supported branches including 22.x, 24.x, and 26.x prior to the October 21, 2026 security release. Patched versions 22.23.4, 24.22.1, and 26.12.1 address the memory corruption issue that arises during Brotli decompression in services accepting third-party content. The same update resolves additional high-severity problems, notably hostname validation weaknesses in TLS. Organizations using unsupported branches such as 18.x, 20.x, 23.x, and 25.x must migrate to maintained releases, while those linking against system Brotli libraries should verify inclusion of upstream commit 0ed4268. Security teams are advised to audit all environments, including containers and CI pipelines, that process Brotli-encoded data.
PHP Issues Emergency Patches for Critical PHP2Shell RCE Flaws as Exploit Attempts Emerge
PHP has released urgent patches for two critical vulnerabilities collectively known as PHP2Shell that enable unauthenticated remote code execution on affected servers. The flaws, tracked as CVE-2026-109618 and CVE-2026-109619, score 9.2 on CVSS v4.0 and have already seen active exploitation attempts with malicious payloads adjusted to bypass WAF rules. The issues affect multiple PHP deployment methods including Apache with mod_php, PHP-FPM, php-cgi, FrankenPHP, and LiteSpeed, and do not require complex application code to be exploitable. CVE-2026-109618 involves a memory corruption chain leading to code execution under the web process privileges, while CVE-2026-109619 stems from an obsolete global state issue causing use-after-free conditions that can result in memory disclosure and code execution. Patched versions include PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35, with administrators advised to restart all persistent workers after updating. Temporary mitigations such as setting memory_limit to -1 and disabling session.upload_progress.enabled are available for environments that cannot patch immediately.
Cyberattack Exploits Exposed Danfoss Refrigeration Controllers Across French Supermarkets
A remote cyberattack targeted Danfoss AK-SM 8xxA Series controllers in French supermarkets, allowing attackers to manipulate refrigeration parameters and trigger defrost cycles that raised temperatures and spoiled food stock. The incident impacted an estimated 250 to 350 stores belonging to major chains including Intermarché, Carrefour Market, Leclerc, and others. Paris prosecutors have opened a large-scale investigation into the attack and associated damages, including potential fires. The root cause traces to outdated software on internet-exposed OT management interfaces, with multiple CVEs enabling authentication bypass and remote code execution. Thousands of similar Danfoss systems remain publicly accessible worldwide, highlighting persistent OT exposure risks. Immediate remediation steps include closing web consoles, enforcing VPN access, and applying updates to versions 4.2 and 4.3.1 or later.
Attackers Exploit Critical Atlassian Data Center Flaw CVE-2026-21589 Hours After PoC Release
Exploitation attempts against CVE-2026-21589 began almost immediately after technical details and a Nuclei template were published. The vulnerability allows unauthenticated arbitrary file read in multiple Atlassian Data Center products and carries a CVSS v4.0 score of 9.3. In environments integrated with Crowd, attackers who obtain crowd.properties can extract plaintext credentials and escalate to administrator privileges via the Crowd API. The flaw stems from improper handling of double-colon sequences in a shared web resource library, enabling path traversal against plugin resource endpoints. Affected products include Bitbucket Data Center, Confluence Data Center, Jira Software Data Center, Jira Service Management Data Center, Bamboo Data Center, Crowd Data Center, Crucible, and Fisheye. Atlassian urges immediate patching outside normal cycles and recommends WAF rules or Tomcat RewriteValve configurations to block traversal patterns. Organizations should also review access logs for double-decoded URLs containing .., /, \, or :: sequences.
LibreOffice and Apache OpenOffice Flaws Enable Remote Code Execution via Malicious Spreadsheets Without Macro Warnings
Two vulnerabilities, CVE-2026-63277 in LibreOffice Calc and CVE-2026-59265 in Apache OpenOffice, allow attackers to execute arbitrary code simply by tricking users into opening specially crafted spreadsheet files. The flaws exploit Java integration and class path handling, bypassing traditional macro security prompts entirely. LibreOffice has already released fixes in versions 26.2.5 and 26.8.0 that restrict class path entries to local file URLs only. Apache OpenOffice 4.1.16 and earlier remain vulnerable, with the stable patch expected in 4.1.17; interim mitigation requires disabling Java integration. The issues highlight risks in office suites that process untrusted documents containing external data connections or JDBC references. Organizations are advised to enforce least-privilege execution and avoid opening files from unknown sources until patches are applied.