Researchers uncovered fast16, a cyber sabotage framework from 2005 that predates Stuxnet by five years. The toolset includes fast16.sys, a kernel driver that selectively targets high-precision calculation software by patching code in memory to corrupt computational results. Combined with self-propagation mechanisms via a Lua-powered carrier module (svcmgmt.exe), the framework spreads across facilities to produce consistent inaccurate calculations. This operation represents the first documented instance of strategic cyber sabotage targeting ultra-expensive computing workloads in advanced physics, cryptographic, and nuclear research. The framework uses an embedded Lua virtual machine predating Flame by three years and appears in the ShadowBrokers leak of NSA Territorial Dispute components with the evasion signature: 'fast16 *** Nothing to see here – carry on ***'.
A malicious Chrome extension impersonating Google's Authenticator application has been identified as part of an ongoing campaign active since early 2026. The extension requests excessive permissions and contains dormant infrastructure suggesting a staged deployment model where malicious updates can be delivered without requiring further user approval. This extension is linked to at least six others through a shared developer front, with two already carrying fully operational malicious payloads. These extensions utilize hidden iframes to inject attacker-controlled content, deploy fraudulent paywalls for free services, and maintain bidirectional communication with command and control servers. The infrastructure maps directly to the AiFrame campaign, which has reportedly compromised over 260,000 users from 2025 to present, marking a continued evolution of this threat.
Cybercriminals are merging traditional malware operations with cryptocurrency-focused attacks, creating hybrid threat ecosystems. Modern crypto drainers have evolved into automated systems capable of extracting assets across multiple blockchains with minimal user interaction, supported by well-developed underground marketplaces offering drainer-as-a-service kits. Two case studies exemplify this convergence: StepDrainer operates as a multichain drainer-as-a-service platform that abuses Web3Modal and smart contract methods across over 20 blockchain networks, using AI-themed lures and polished interfaces to deceive victims into connecting wallets. EtherRAT represents a hybrid Windows implant delivered through trojanized TFTP installers, combining traditional RAT capabilities with blockchain-aware functionality including Ethereum RPC endpoints and embedded wallet addresses. Both threats demonstrate how cryptocurrency theft infrastructure now intersects with mainstream attack surfaces affecting enterprise envir...
Google Threat Intelligence Group identified a sophisticated intrusion campaign by UNC6692 that combined persistent social engineering with custom malware. The attackers impersonated IT helpdesk personnel via Microsoft Teams, leveraging initial email spam campaigns to create urgency. Victims were tricked into downloading AutoHotKey scripts that installed SNOWBELT, a malicious browser extension establishing persistence through scheduled tasks. The modular SNOW ecosystem enabled deep network penetration: SNOWBELT provided initial access, SNOWGLAZE created encrypted WebSocket tunnels masking traffic as legitimate cloud communications, and SNOWBASIN functioned as a local backdoor for command execution. UNC6692 performed internal reconnaissance, escalated privileges by extracting LSASS memory, and used Pass-The-Hash techniques to access domain controllers. The operation culminated in exfiltration of Active Directory databases and credentials via LimeWire, demonstrating advanced tradecraft abusing legitimate clou...
Threat actors are leveraging fake CAPTCHA pages to trick victims into sending premium SMS messages as part of an international revenue share fraud (IRSF) scheme. Operating since at least June 2020, this campaign uses traffic distribution systems and social engineering to direct users through multi-stage fake verifications requiring SMS messages to international phone numbers across 17 countries with high termination fees. Each CAPTCHA step triggers messages to over a dozen destinations, generating over 60 SMS messages per victim costing approximately $30. The operation employs back button hijacking, sophisticated tracking cookies, and affiliate advertising networks to maximize reach while obscuring the fraud from detection. Both individual victims and telecommunication carriers suffer financial losses through this deceptive scheme.
ESET researchers discovered a previously undocumented China-aligned APT group named GopherWhisper that targeted a governmental entity in Mongolia. The group employs a diverse arsenal of custom tools, predominantly written in Go, including backdoors LaxGopher, RatGopher, and BoxOfFriends, along with injectors JabGopher, exfiltration tool CompactGopher, loader FriendDelivery, and C++ backdoor SSLORDoor. The threat actors abuse legitimate services including Discord, Slack, Microsoft 365 Outlook, and file.io for command and control communications and data exfiltration. Through extraction of thousands of messages from compromised Slack and Discord channels, researchers gained valuable insights into the group's internal operations and post-compromise activities. Timestamp analysis of communications indicates operators work during UTC+8 business hours, aligning with China Standard Time, supporting attribution to China-aligned actors.
A sophisticated ClickFix campaign was detected in mid-March 2026, beginning with a malicious webpage impersonating Booking.com's visual identity with a fake CAPTCHA. The attack leverages social engineering to trick victims into executing a PowerShell command that downloads and runs a script directly in memory. The JavaScript code automatically copies malicious commands to the clipboard and intercepts copy events. Once executed, the PowerShell dropper performs system fingerprinting, downloads a ZIP payload from a remote server, deploys it to user directories, establishes persistence through registry keys and scheduled tasks, and executes the final payload. The campaign demonstrates well-structured code with fallback mechanisms and real-time telemetry via Telegram, suggesting the use of a ready-to-use attack kit.
DinDoor is a Deno-based backdoor delivered via MSI files that exploits the Deno runtime to execute obfuscated JavaScript for command and control communications and system fingerprinting. Two analyzed samples show different execution behaviors: one writes JavaScript to disk while the other executes entirely in memory. Both samples use identical fingerprinting algorithms generating unique victim identifiers. One sample contains an embedded JWT exposing campaign metadata and the domain serialmenot[.]com, identified as multi-tenant infrastructure serving multiple threat actors including state-sponsored groups and cybercriminals. Analysis of HTTP response headers enabled identification of 20 active C2 servers across 15 autonomous systems, many using bulletproof hosting providers. The malicious infrastructure uses Caddy proxy with distinctive headers allowing network-based detection.
Attackers are leveraging the trusted reputation of Foxit PDF Reader, used by over 650 million people, to distribute malicious installers disguised as legitimate software. Rather than exploiting vulnerabilities, threat actors impersonate the vendor through fake installers with document-themed filenames that bypass user suspicion. When executed, these files display decoy passport images while downloading malicious MSI packages that deploy UltraVNC remote access tools disguised as GPU drivers. The attack establishes persistence through registry modifications and firewall exceptions, connecting to attacker-controlled infrastructure for complete remote system control. Telemetry indicates broad distribution across Germany, the United States, the United Kingdom, and Ukraine. This campaign demonstrates how brand impersonation combined with social engineering proves more effective than technical exploits, relying on user trust and behavioral patterns rather than software vulnerabilities.
X-Labs researchers discovered 10 verified Indirect Prompt Injection (IPI) payloads deployed across live web infrastructure. Unlike direct prompt injection where users send malicious input to AI models, IPI hides adversarial instructions inside ordinary web content. When AI agents crawl or summarize poisoned pages, they ingest and execute these instructions as legitimate commands. The discovered payloads span financial fraud, data destruction, API key exfiltration, and denial-of-service attacks. Attackers employ techniques including CSS invisibility, HTML comments, accessibility attribute abuse, meta namespace spoofing, and system prompt tag impersonation. The shared injection templates across multiple domains suggest organized tooling rather than isolated experimentation. Observed attack intents include unauthorized financial transactions, terminal command execution, content suppression, traffic hijacking, and sensitive information leakage, targeting AI systems that browse web pages, index content for RAG ...
On March 12, 2026, a sophisticated attack campaign was identified targeting Chinese-speaking individuals using military-themed document lures distributed through a malicious ZIP archive. The operation employed a trojanized SumatraPDF binary as the initial vector to deploy an AdaptixC2 Beacon and Visual Studio Code on victim systems. The shellcode loader demonstrated significant similarities to the TOSHIS loader previously linked to TAOTH campaigns. Attackers established a custom AdaptixC2 Beacon listener utilizing GitHub for command-and-control infrastructure. The staging server infrastructure additionally hosted CobaltStrike Beacon and EntryShell backdoor, both previously associated with this threat group. The campaign infrastructure included multiple compromised domains and IP addresses for malware distribution and C2 communications.
A phishing campaign utilized a fraudulent Adobe-themed website to trick victims into downloading and executing ScreenConnect remote access software. Once initial access was established, threat actors conducted interactive operations deploying multiple malicious binaries including a credential harvesting tool named password.exe. The attackers also exploited the ms-phone URI handler to launch the Phone Link application, attempting to socially engineer victims into linking their mobile devices to potentially capture notifications, authentication prompts, and sensitive information. The attack demonstrates a multi-stage compromise focusing on persistence establishment, credential theft, and preparation for potential lateral movement across the victim's network infrastructure.
Investigation of DPRK-linked fake IT worker infrastructure began after cryptocurrency researcher ZachXBT identified domain luckyguys[.]site connected to illicit payments. Analysis of 30 days of network activity associated with IP 163.245.219[.]19 revealed concentrated VPN usage patterns, with Astrill VPN (37.5%), Mullvad (32.25%), and Proton VPN (6.25%) being prominent. American and Latvian residential IPs communicated with the infrastructure, showing frequent Astrill VPN usage and connectivity to Gmail, ChatGPT, and Workana freelance platform. A second IP, 216.158.225[.]144, was discovered through X509 certificate analysis. Traffic dropped sharply following public exposure, consistent with adversary behavior of abandoning attributed infrastructure. The activity suggests a distributed network of remote IT workers participating in sanctions evasion workflows, leveraging AI tools and freelance platforms to obtain employment under false identities.
This analysis documents a newly observed ClickFix variant that abuses native Windows utilities, specifically cmdkey and regsvr32, for payload delivery. Victims are socially engineered through fake CAPTCHA pages to execute a malicious command via the Windows Run dialog. The single command chains multiple actions: staging credentials using cmdkey, retrieving a remote DLL via regsvr32 from a UNC path, and executing it silently. The 64-bit DLL establishes persistence through a scheduled task pulled from a remote XML file hosted on attacker infrastructure. This approach avoids traditional malware drops and leverages exclusively trusted Windows components for high stealth. The variant demonstrates continued evolution of ClickFix techniques, moving beyond PowerShell to use command chaining with legitimate system tools.
Docker and Socket uncovered a supply chain compromise affecting Checkmarx KICS distribution channels. Attackers poisoned official Docker Hub images (tags v2.1.20, v2.1.21, alpine) and VS Code extensions (versions 1.17.0, 1.19.0), introducing unauthorized data exfiltration capabilities. The trojanized KICS binary collects and encrypts scan reports containing credentials from infrastructure-as-code files, transmitting them to external endpoints. Compromised VS Code extensions download mcpAddon.js via Bun runtime, harvesting GitHub tokens, AWS credentials, Azure tokens, npm configurations, and SSH keys. The malware creates public GitHub repositories for staging stolen data, injects malicious GitHub Actions workflows to capture repository secrets, and uses stolen npm credentials to identify writable packages for propagation. TeamPCP appears to claim responsibility for this multi-stage attack designed to steal developer credentials and propagate through CI/CD pipelines.
Malicious npm packages associated with Namastex.ai were compromised with malware exhibiting tradecraft similar to TeamPCP's CanisterWorm campaign. The attack targeted packages including @automagik/genie and pgserve, implementing install-time execution that harvests credentials, environment variables, SSH keys, cloud credentials, browser data, and crypto-wallet artifacts. The payload exfiltrates stolen data to both a conventional webhook at telemetry.api-monitor.com and an Internet Computer Protocol canister endpoint. It incorporates self-propagation logic to compromise additional npm packages using stolen publishing tokens and includes cross-ecosystem spreading capabilities targeting PyPI. The malware uses hybrid encryption with RSA and AES-256-CBC for data exfiltration. Multiple package namespaces were affected, suggesting shared infrastructure or coordinated compromise across publisher accounts.
FudCrypt is a Cryptor-as-a-Service platform offering subscription-based malware obfuscation for $800 to $2,000 monthly. The service wraps customer payloads in multi-stage deployment packages featuring DLL sideloading, AMSI and ETW interference, silent UAC elevation via CMSTPLUA, and Windows Defender tampering through Group Policy. Analysis of recovered server infrastructure revealed 200 registered users, 334 builds, and comprehensive fleet C2 command history across 32 enrolled agents. The operator maintains a separate signing infrastructure using four Azure Trusted Signing accounts to sign operator-controlled binaries including fleet agents, native loaders, and ScreenConnect installers. The platform employs 20 undocumented DLL sideload carrier profiles, per-build polymorphic encryption with layered XOR-32, RC4-16, and custom S-box transforms, and an advanced development branch featuring indirect syscalls, module stomping, fiber-based execution, and Ekko sleep obfuscation. Server infrastructure included exp...
Two distinct phishing campaigns have been identified targeting companies in Greece, Spain, Slovenia, Bosnia and Central American countries to deliver FormBook data-stealing malware. The first campaign uses RAR attachments containing legitimate executables like Sandboxie ImBox.exe, TikTok desktop, Adobe PDF Preview Handler, and XZ Utils, exploiting DLL side-loading with malicious DLL files. The second campaign deploys heavily obfuscated JavaScript that drops encrypted PNG files, uses PowerShell with Base64 encoding, and leverages a custom .NET loader called Mandark to inject the payload into RegAsm process. Both campaigns deliver the same FormBook executable that employs advanced evasion by manually mapping ntdll.dll in memory to bypass user-mode monitoring and perform direct syscalls, enabling credential theft and data collection from browsers while avoiding detection mechanisms.
A sophisticated multi-stage malware operation was identified through an exposed C2 panel at 103.241.66[.]238:1337, combining cryptocurrency clipboard hijacking across eight chains, BIP-39 seed phrase theft, browser credential exfiltration, ransomware module (crpx0), and Java RAT builder managed via FastAPI-based panel with license key system. The operation targets Windows and macOS using FedEx and OnlyFans-themed social engineering lures, with complete source code exposed in open directories. The ransomware component communicates with three Russian .ru domains resolving to 31.31.198[.]206 at REG.RU hosting, operating under the identity DataBreachPlus with Telegram, qTox, and ProtonMail contacts. Ten cryptocurrency wallet addresses spanning Bitcoin, Ethereum, Tron, Dogecoin, Litecoin, Solana, Ripple, and Bitcoin Cash were extracted from configurations, indicating a Malware-as-a-Service operation with tiered licensing.
Kyber ransomware represents a significant threat through dual-platform deployment capabilities targeting VMware ESXi virtualization infrastructure and Windows file systems. During a March 2026 incident response engagement, two Kyber payloads were recovered from the same environment. The ESXi variant, written in C++, specifically targets VMware environments with datastore encryption, VM termination, and management interface defacement capabilities. The Windows variant, written in Rust, includes experimental Hyper-V targeting features. Both samples share campaign identifiers and Tor-based infrastructure, confirming coordinated cross-platform operations. Despite advertising post-quantum Kyber1024 encryption, the ESXi variant actually uses ChaCha8 with RSA-4096 key wrapping, while the Windows variant implements the claimed AES-256-CTR with Kyber1024 hybrid scheme. The ransomware includes anti-recovery measures, service termination, and effective encryption strategies designed to cause complete operational disr...
