A sophisticated 64-bit information-stealing malware named Remus has emerged as a direct evolution of the notorious Lumma Stealer. Following the doxxing of alleged Lumma core members between August and October 2025, developers created this advanced variant, with test builds appearing in September 2025 and live campaigns starting February 2026. Remus employs innovative techniques including injecting custom 51-byte shellcode into browser memory to extract protected master keys, bypassing Application-Bound Encryption in Chromium-based browsers. The malware utilizes EtherHiding through Ethereum smart contracts for command-and-control resolution, making infrastructure takedowns nearly impossible. It targets browser credentials, session cookies, and cryptocurrency wallets while implementing rigorous anti-analysis checks to evade security research environments.
A sophisticated credential theft campaign targeting over 35,000 users across 13,000 organizations was observed between April 14-16, 2026. The operation primarily impacted the United States, particularly healthcare and financial services sectors. Attackers used code of conduct themed phishing emails masquerading as internal compliance communications, sent through legitimate email delivery services from attacker-controlled domains. Victims received polished HTML emails with PDF attachments containing fake disciplinary logs and CAPTCHA gates to evade automated analysis. The multi-stage attack chain ultimately directed users to counterfeit Microsoft authentication pages operating as adversary-in-the-middle infrastructure, enabling real-time interception of credentials and session tokens while bypassing multi-factor authentication defenses.
An exposed command and control server on RouterHosting infrastructure revealed an active Iranian-nexus intrusion campaign targeting twelve Omani government ministries. The operation primarily focused on the Ministry of Justice and Legal Affairs, deploying custom webshells that provided persistent access through April 2026. Over 26,000 user records containing judicial case data, committee decisions, and registry hives were exfiltrated. The attacker utilized ProxyShell exploits, DotNetNuke vulnerabilities, and custom Python scripts targeting Exchange servers, SQL databases, and Oracle systems. Infrastructure analysis revealed connections to spoofed Iranian diaspora media and censorship circumvention tools, with tactical overlaps indicating MOIS-linked groups such as APT34 and MuddyWater. The campaign specifically targeted judicial records, immigration systems, and citizen identity data across multiple government entities.
In March 2026, threat actors weaponized the OpenClaw AI agent framework by publishing a deceptive "DeepSeek-Claw" skill. This skill embedded malicious installation instructions designed to trick AI agents and developers into executing hidden payloads. On Windows systems, a PowerShell command downloads an MSI package containing a legitimate signed GoToMeeting executable that sideloads a malicious DLL. This loader patches ETW and AMSI for evasion, then decrypts and executes Remcos RAT using TEA encryption, enabling remote access and data theft including keylogging and cookie stealing. An alternate execution path for macOS and Linux delivers GhostLoader through obfuscated Node.js scripts, harvesting credentials via fake sudo prompts and exfiltrating SSH keys, cryptocurrency wallets, and cloud API tokens. This campaign represents an emerging threat vector exploiting autonomous AI workflows and developer trust in open-source frameworks.
Since April 8, 2026, installers of DAEMON Tools software have been compromised with malicious payloads distributed through the legitimate website. Versions 12.5.0.2421 to 12.5.0.2434 contain trojaned binaries (DTHelper.exe, DiscSoftBusServiceLite.exe, DTShellHlp.exe) signed with legitimate developer certificates. The attack has affected thousands of systems across over 100 countries, though advanced payloads were selectively deployed to approximately a dozen machines in government, scientific, manufacturing, and retail organizations. Initial infection establishes backdoor communications to typosquatted domains, followed by deployment of an information collector for system profiling. Targeted systems receive additional implants including a minimalistic backdoor and QUIC RAT. Chinese-language strings found in malicious components suggest a Chinese-speaking threat actor. The attack remains active at time of publication, demonstrating sophisticated supply chain compromise techniques comparable to the 2023 3CX ...
Cisco Talos uncovered an intrusion active since January 2026 where attackers deployed CloudZ remote access tool and an undocumented plugin called Pheno to steal credentials and one-time passwords. The attack exploits Microsoft Phone Link application by intercepting synchronized mobile data including SMS and OTPs without requiring phone-level infection. CloudZ evades detection through dynamic memory execution and anti-analysis checks. The infection chain begins with a fake ScreenConnect update executable, leading to a Rust-compiled dropper that deploys a .NET loader, ultimately establishing the modular CloudZ RAT. The Pheno plugin monitors Phone Link processes and intercepts SQLite database files containing synchronized phone data. CloudZ employs ConfuserEx obfuscation, multiple configuration layers, and facilitates various commands including browser data exfiltration, shell execution, and plugin management while maintaining persistence through scheduled tasks.
North Korea-aligned APT group ScarCruft executed a multiplatform supply-chain attack targeting ethnic Koreans in China's Yanbian region, an area significant for North Korean refugees and defectors. Since late 2024, the group compromised a video gaming platform dedicated to Yanbian-themed games, trojanizing both Windows and Android components with the BirdCall backdoor. The Windows client received malicious updates leading to RokRAT and subsequently BirdCall deployment, while Android games were directly trojanized. This marks the first discovery of Android BirdCall, capable of comprehensive surveillance including data collection, screenshots, and voice recording. The campaign focuses on espionage against individuals of interest to the North Korean regime, particularly refugees and defectors.
An emerging threat group is conducting a global SEO-poisoning campaign distributing trojanized Microsoft Teams installers that deploy a multi-stage shellcode loader and backdoor designated Lorem Ipsum. Active since February 2026, the campaign targets users searching for Microsoft Teams across six countries, with confirmed targeting of a US healthcare organization. The operators evolved rapidly from minimally obfuscated test builds to sophisticated loaders featuring substitution cipher decoding, XOR-encrypted shellcode, DLL sideloading, and JFIF-disguised C2 traffic. The malware distinctively abuses letsdiskuss[.]com, a legitimate India-based platform, as a dead-drop resolver for C2 infrastructure. Attackers use validly signed MSI installers with three-day Microsoft ID Verified certificates, NameCheap-registered infrastructure weaponized within hours, and per-victim UUID-tracked callbacks. Development velocity suggests possible LLM-assisted tooling, indicating a well-funded mid-tier criminal actor operating...
KATARU is an IoT malware variant identified in 2026 that uses Telnet credential brute-forcing to infect devices. It combines traditional Mirai-style botnet features with multiple Linux local privilege escalation exploits and advanced persistence mechanisms across Linux and embedded platforms. The malware uses encrypted command and control communications and employs anti-analysis and decoy traffic techniques. Artifacts suggest AI-assisted development, including mismatched shellcode and untested cross-platform logic. It supports various DDoS attack methods and SSH brute-forcing.
The Harvester APT group has developed a new Linux version of its GoGra backdoor that uses the legitimate Microsoft Graph API and Outlook mailboxes as a covert command-and-control channel. The malware employs social engineering lures with tailored decoy documents, masquerading malicious ELF files as standard documents. Initial VirusTotal submissions originated from India and Afghanistan, indicating these regions as primary targets. The backdoor uses hardcoded Azure AD credentials to poll a specific mailbox folder at two-second intervals, executing commands received via encrypted emails and exfiltrating results through reply messages. Analysis confirms this Linux variant shares nearly identical code with a previously known Windows version, including matching spelling errors, demonstrating Harvester's multi-platform development strategy and continued focus on South Asian espionage operations.
A malicious artifact of the widely-used intercom/intercom-php package version 5.0.2 was discovered on Packagist, representing an expansion of the Mini Shai-Hulud supply chain attack from npm into the PHP ecosystem. The compromised package exploits Composer plugin execution to download Bun runtime and execute an obfuscated credential-stealing payload during installation. The malicious code harvests sensitive credentials including GitHub tokens, cloud provider credentials, SSH keys, Kubernetes tokens, and HashiCorp Vault secrets from developer machines and CI/CD environments. Stolen data is encrypted using AES-256-GCM and exfiltrated to attacker-controlled infrastructure. The payload also contains propagation logic to modify GitHub repositories and npm packages using stolen credentials. With approximately 12,700 daily installs, the compromised artifact potentially reached numerous high-value development environments before removal.
The intercom-client npm package version 7.0.4 was compromised through a malicious GitHub account, introducing credential-stealing malware into a widely used Node.js SDK with approximately 360,000 weekly downloads. The attack deployed two malicious files: setup.mjs, executed via preinstall hook to download an unverified Bun binary, and router_runtime.js, an obfuscated 11.7 MB script targeting Kubernetes, Vault, and cloud credentials. Stolen data was encrypted and exfiltrated through GitHub API. The compromise resembles recent attacks on PyPI lightning package and SAP CAP packages, sharing technical patterns with TeamPCP-linked campaigns including GitHub-based exfiltration and CI/CD targeting. The attack was facilitated by compromised GitHub account nhur, which created malicious workflows and triggered automated CI publishing, affecting developers and CI/CD environments that installed the package.
Vect ransomware emerged in January 2026 as a new threat actor operating a Ransomware-as-a-Service program with strategic partnerships that significantly expand its reach. The group has partnered with TeamPCP, known for supply chain attacks compromising security tools like Trivy, KICS, and LiteLLM, and BreachForums, distributing affiliate keys to forum members. With 25 published victims primarily targeting the United States and Technology sector, Vect maintains an open affiliate program requiring only a $250 invite code. The operation offers multi-platform ransomware payloads for Windows, Linux, and ESXi with sophisticated lateral movement capabilities and tiered commission structures reaching 89% for top affiliates. Analysis reveals connections to the defunct Devman ransomware through shared code strings and ransom note similarities, suggesting possible rebranding or code reuse.
The popular PyPI package lightning experienced a supply chain attack affecting versions 2.6.2 and 2.6.3, published on April 30, 2026. The compromise introduced malicious code that executes automatically upon module import, downloading Bun JavaScript runtime and executing an 11MB obfuscated payload. The attack harvests credentials including GitHub tokens, npm tokens, cloud credentials from AWS, Azure, and Google Cloud, while targeting CI/CD environments. The malicious code poisons GitHub repositories by injecting backdoored files impersonating Claude Code commits and infects local npm packages through tarball manipulation. The attack shows similarities to previous Shai-Hulud campaigns in terms of credential targeting and obfuscation methods. Evidence suggests the maintainer's GitHub account (pl-ghost) was compromised, with suspicious branch operations and disclosure suppression indicating ongoing attacker control. The incident affects a widely-used deep learning framework receiving millions of monthly downl...
BackgroundFix masquerades as a free image-editing tool but functions as a ClickFix social engineering lure. The fake service prompts users to verify they are human, copying malicious commands to their clipboard that invoke finger.exe to retrieve additional payloads. This chain delivers CastleLoader, which subsequently drops NetSupport RAT and a custom .NET stealer dubbed CastleStealer. The loader uses reflective PE injection, API hashing, and ChaCha20-encrypted C2 communications. CastleStealer targets browser credentials, cryptocurrency wallet extensions, and Telegram sessions through DPAPI decryption and Restart Manager APIs. The campaign leverages BYOI tactics with embedded Python interpreters and multiple shellcode stages. A notable implementation flaw exists where launch method 4 references regsrv32.exe instead of the correct regsvr32.exe, causing silent failures.
KarstoRAT is a newly identified remote access trojan that emerged in early 2026, combining surveillance, credential theft, and remote command execution capabilities. The malware supports extensive post-compromise operations including system reconnaissance, screenshot and audio capture, webcam monitoring, keylogging, and token theft. It communicates with a C2 server at 212.227.65[.]132 using HTTP protocols with the user agent 'SecurityNotifier'. Distribution occurs through gaming-themed lure pages targeting Roblox players and FPS/GTA modders via fake cheat loaders. KarstoRAT employs multiple persistence mechanisms through registry keys, scheduled tasks, and startup folders, while featuring a UAC bypass using the fodhelper.exe technique. The malware has not been publicly advertised on cybercrime forums, suggesting private development and limited operator use rather than commodity distribution.
MioLab, also known as Nova, is a sophisticated Malware-as-a-Service platform targeting macOS environments, heavily advertised on Russian-speaking underground forums. The platform features extensive data exfiltration capabilities, including browser credential theft, cryptocurrency wallet targeting (supporting over 200 browser extensions and 50+ desktop wallets), and a premium module specifically designed to compromise Ledger and Trezor hardware wallets by intercepting 24-word BIP39 recovery seed phrases. The lightweight C-based payload supports both Intel and Apple Silicon architectures across macOS versions from Sierra to Tahoe. MioLab employs sophisticated social engineering through customizable DMG builders with live preview features, fake system prompts, and ClickFix integration. Recent updates demonstrate rapid development, including Safari cookie grabbing, automated Apple Notes decryption, and universal hardware wallet modules. The operation utilizes bulletproof hosting services and shares infrastruct...
Multiple npm packages in the SAP JavaScript and cloud application development ecosystem were compromised in a suspected supply chain attack. Affected packages include mbt@1.2.48, @cap-js/db-service@2.10.1, @cap-js/postgres@2.2.2, and @cap-js/sqlite@2.2.2. The compromised versions introduced malicious preinstall scripts that download and execute Bun binaries from GitHub, then run heavily obfuscated payloads designed to harvest credentials from developer machines and CI/CD environments. The payloads steal SSH keys, cloud credentials, npm tokens, GitHub access, cryptocurrency wallets, and CI/CD secrets directly from runner memory. Stolen data is encrypted and exfiltrated via GitHub repositories created under victim accounts. The malware also attempts self-propagation by injecting itself into additional packages using stolen npm tokens and establishes persistence through VSCode and Claude IDE configurations.
A supply chain operation dubbed 'Mini Shai Hulud' compromised SAP-related npm packages by injecting malicious preinstall scripts that execute during installation. The campaign leverages multi-stage payloads to harvest developer and CI/CD secrets from GitHub, npm, and major cloud providers, exfiltrating data via attacker-controlled GitHub repositories. Malicious versions of legitimate SAP ecosystem packages execute obfuscated payloads that collect GitHub tokens, npm credentials, cloud secrets from AWS, Azure and GCP, Kubernetes tokens, and GitHub Actions secrets. The malware includes propagation logic to infect additional repositories and features browser credential theft capabilities. It performs language checks to avoid Russian-speaking systems. Attribution points to TeamPCP based on shared RSA public keys and overlapping techniques from previous operations.
On April 16, 2026, a threat actor leveraged stolen VPN credentials to access a Windows workstation and deployed a SYSTEM-level backdoor using the Komari agent, an open-source monitoring tool with built-in command-and-control capabilities. The attacker authenticated through an SSLVPN session from IP 45.153.34[.]132 and used Impacket smbexec.py to enable RDP on the target system. The Komari agent was installed as a persistent Windows service named 'Windows Update Service' using NSSM, pulling the installer directly from the official GitHub repository. Komari provides bidirectional control through WebSocket connections, offering arbitrary command execution, interactive reverse shell access, and network probing capabilities by default. Microsoft Defender quarantined an earlier registry dump attempt, forcing the adversary to pivot to this GitHub-based approach. This represents the first publicly documented case of Komari being abused in a real-world intrusion.
