CVE-2026-39987 is a critical pre-authorization remote code execution vulnerability in the Marimo Python notebook platform. Shortly after disclosure, multiple threat actors exploited this flaw to deploy malware via typosquatted HuggingFace Spaces, leveraging the NKN blockchain for command and control. The malware, disguised as a Kubernetes tool named kagent, established persistence on infected systems and targeted AWS credentials, API tokens, and databases such as PostgreSQL and Redis. Exploitation events were observed from multiple countries, demonstrating active and widespread abuse of this vulnerability in AI/ML infrastructure environments.
A sophisticated phishing campaign is targeting YouTube creators using convincing fake copyright strike notifications. The attack dynamically pulls real channel data including profile pictures, subscriber counts, and recent videos to create personalized scare pages. Victims are funneled through a Browser-in-the-Browser attack displaying a fake Google sign-in that captures credentials. The operation functions as phishing-as-a-service, with multiple attackers sharing infrastructure and rotating domains to evade detection. Successful attacks result in complete Google account takeover, allowing hijackers to rebrand channels and livestream cryptocurrency scams to existing audiences. The kit automatically exempts channels with over three million subscribers to avoid detection by security teams.
A threat cluster has been identified leveraging a customized Adwind (Java RAT) variant with polymorphic characteristics to deliver JanaWare ransomware. The campaign specifically targets Turkish users through geofencing mechanisms that check system locale and external IP geolocation. Active since at least 2020, the operation primarily affects home users and small to medium-sized businesses. Initial access occurs via phishing emails with malicious Java archives distributed through Google Drive links. The ransomware employs AES encryption and communicates over Tor networks, demanding modest ransoms between $200-$400. The malware uses multiple obfuscation techniques including Stringer and Allatori obfuscators, implements file pumping for polymorphism, and disables Windows security features before encryption. Victims are instructed to contact attackers through qTox or dedicated Tor onion sites.
A sophisticated macOS stealer called notnullOSX emerged in March 2026, developed by threat actor alh1mik (formerly 0xFFF) who returned after a 2023 exit from underground forums. This Go-written modular stealer exclusively targets macOS users with cryptocurrency holdings exceeding $10,000. Distribution occurs through ClickFix social engineering and malicious DMG files disguised as legitimate applications like WallSpace. The malware employs a modular architecture with specialized components to exfiltrate iMessage history, Apple Notes, browser credentials, Safari cookies, crypto wallet files, SSH keys, and cloud provider credentials. By social-engineering victims into granting Full Disk Access, notnullOSX bypasses macOS TCC protections without triggering permission dialogs. The stealer maintains persistent WebSocket connections to Firebase infrastructure, functioning as both an infostealer and backdoor with remote module update capabilities.
In spring 2022, a highly sophisticated backdoor named AngrySpark was discovered on a single machine in the United Kingdom. The malware employed a three-stage architecture: a DLL masquerading as a Windows Task Scheduler component, a custom virtual machine interpreter running bytecode instructions, and a beacon that profiles systems while disguising C2 communications as PNG image requests. The malware featured VM-based obfuscation, dual encrypted C2 channels using RSA-4096 and XXTEA encryption, direct syscalls bypassing usermode hooks, hypervisor detection, and CET-aware anti-analysis capabilities. It operated for approximately one year with active maintenance visible through syscall table updates and configuration changes between May 2022 and January 2023. The infrastructure expired in mid-2023 and the operation ceased, with no additional samples or victims identified despite the significant engineering effort invested in its development.
A coordinated campaign of 108 malicious Chrome extensions operated through shared command-and-control infrastructure at cloudapi[.]stream has been identified, collectively accounting for approximately 20,000 installations. The campaign spans multiple threat categories: 54 extensions steal Google account identities via OAuth2, one extension actively exfiltrates Telegram Web sessions every 15 seconds, and 45 extensions contain a universal backdoor enabling arbitrary URL execution on browser startup. Published under five distinct publisher identities (Yana Project, GameGen, SideGames, Rodeo Games, and InterAlt), these extensions masquerade as legitimate tools including Telegram sidebar clients, slot games, YouTube and TikTok enhancers, and translation utilities. All extensions route stolen credentials, user identities, and browsing data to servers controlled by the same operator, with infrastructure confirming a Malware-as-a-Service business model.
A Brazilian banking fraud operation leveraging ClickFix social engineering was discovered through a community tip, exposing a completely unauthenticated command-and-control infrastructure. The campaign deploys a malicious Chrome extension masquerading as a Banco Central do Brasil tool, force-installed via Chrome Cloud Management enrollment tokens. The extension achieves zero antivirus detections while targeting eight Brazilian financial institutions. At investigation time, 59 machines were compromised with seven active connections. The operator's C2 server exposed all endpoints without authentication, including admin panels, live victim screenshots, stolen credentials in cleartext, and intercepted Pix payment data. Attribution was established through WHOIS records revealing the operator's real name, CPF, and email address. The operation specifically targeted Northern Brazilian regional banks and credit cooperatives, with evidence of compromising a school fund account.
APT37 conducted a sophisticated social engineering campaign utilizing Facebook accounts claiming locations in Pyongyang and Pyongsong, North Korea, to conduct reconnaissance and build trust with targets. After establishing relationships through Facebook Messenger, the threat actor migrated conversations to Telegram and employed pretexting tactics, claiming to share encrypted PDF documents containing military weapons information. Victims were persuaded to install a tampered Wondershare PDFelement installer that executed embedded shellcode for initial compromise. The attack chain delivered follow-on commands through a JPG-disguised payload hosted on a compromised Japanese real estate website. The malware abused Zoho WorkDrive OAuth2 APIs as C2 channels, exfiltrating screenshots, documents, system information, and audio files. The campaign employed multiple evasion techniques including code cave injection, process hollowing into legitimate dism.exe, XOR encryption layers, and fileless in-memory execution.
Analysis of attacks against Linux SSH servers during Q1 2026 reveals P2PInfect worm as the dominant threat, representing 70.3% of all attack sources. DDoS botnets including Mirai, XMRig, Prometei, and CoinMiner were identified as primary threats. A notable campaign involved installing V2Ray proxy tools on compromised systems, attributed to a suspected Chinese threat actor. Attackers employed SSH brute-force techniques to gain access, executed reconnaissance commands to assess system information, and deployed V2Ray for proxy node operations. The campaign targeted poorly secured SSH servers with weak credentials, emphasizing the need for strong password policies, access controls, and network monitoring to detect unusual outbound connections and proxy-related activities.
On April 7, 2026, a security researcher described an Adobe Reader zero-day vulnerability that has been exploited since at least December 2025. The vulnerability allows threat actors to execute privileged Acrobat APIs via specially crafted malicious PDF files that execute obfuscated JavaScript when opened. Exploitation allows attackers to steal sensitive user and system data and to potentially launch additional attacks and remotely execute code. Recommendations: Reduce the risk by automatically scanning PDF email attachments, blocking suspicious files, training users to be wary of unsolicited attachments, and advising users to temporarily avoid using Adobe Reader to open PDFs. Reference: https://www.sophos.com/en-us/blog/adobe-reader-zero-day-vulnerability-in-active-exploitation
On April 11, 2026, researchers analyzed a CHM file (api_reference.chm) tagged as Kimsuky that initiated a three-stage attack chain. The C2 server at check[.]nid-log[.]com had directory listing enabled, allowing recovery of complete source code for all payload stages: a 6,338-byte VBScript performing system reconnaissance and establishing persistence via scheduled task, a 449-byte VBScript bridge to PowerShell, and a 6,234-byte PowerShell keylogger with clipboard monitoring and timed exfiltration. The infrastructure included 79+ domains across 5 C2 IPs spanning Korean VPS providers. The server responded with "Million OK !!!!" signature, matching previously documented Kimsuky infrastructure while showing upgraded Apache/PHP stack. The operation targeted Korean Naver users through credential phishing and tax authority impersonation, with infrastructure linked to previously documented Kimsuky campaigns via shared DAOU Technology subnets.
A sophisticated fake recruitment campaign named 'graphalgo' has been active since May 2025, targeting JavaScript and Python developers in the cryptocurrency sector. Attackers approach victims through LinkedIn, Facebook, and Reddit with fabricated job opportunities from fake blockchain companies like Veltrix Capital. The campaign uses malicious dependencies hidden in npm and PyPI packages, delivered through coding test repositories on GitHub. Notable is the bigmathutils package that accumulated over 10,000 downloads before its malicious version was released. The operation deploys a remote access trojan (RAT) with token-protected C2 communication, file manipulation capabilities, and functionality to detect the Metamask browser extension, indicating focus on cryptocurrency theft. The modular campaign design allows threat actors to maintain backend infrastructure while easily replacing compromised frontend elements.
An open directory discovery at refundonex[.]com exposed a complete Phishing-as-a-Service and RAT-as-a-Service platform targeting Spanish and Portuguese-speaking victims. The investigation uncovered 3,788 files including weaponized LNK, VBS, and AES-encrypted PowerShell payloads delivering a remote access trojan. The platform, called Shadow Panel, operates from Bulgarian infrastructure and offers capabilities including remote shell execution, screenshot capture, file management, browser credential theft, clipboard hijacking for cryptocurrency wallets, and multi-operator support. The C2 panel's frontend JavaScript was publicly accessible, revealing 29 API endpoints and the complete architecture. Infrastructure analysis linked the operation to nikola4010@proton[.]me through WHOIS data and historical malicious domain associations dating back to 2021, indicating a long-running cybercriminal operation with minimal detection coverage.
In March 2026, a China-nexus threat actor launched a sophisticated campaign targeting countries in the Arabian Gulf region, exploiting renewed Middle East conflict themes within 24 hours of escalation. The attack utilized Arabic-language lures depicting missile strikes and employed a multi-stage infection chain beginning with weaponized ZIP archives containing malicious LNK and CHM files. The campaign deployed a heavily obfuscated PlugX backdoor variant through DLL sideloading, with components using control flow flattening and mixed boolean arithmetic techniques. The backdoor supports HTTPS command-and-control communications, DNS-over-HTTPS resolution, and multiple plugins for system manipulation. Based on tools, techniques, and procedures including specific RC4 decryption keys and rapid geopolitical weaponization, the activity is attributed with medium confidence to Mustang Panda.
Mirax is a newly identified Android Remote Access Trojan operating as Malware-as-a-Service, actively targeting European users, particularly in Spanish-speaking regions. Distributed through Meta advertisements and GitHub-hosted droppers, the malware has reached over 200,000 accounts. It employs sophisticated techniques including dynamically fetched HTML overlays, comprehensive keylogging, and remote device control capabilities. A distinctive feature is its integration of SOCKS5-based residential proxy functionality, transforming infected devices into proxy nodes that enable attackers to route traffic through legitimate residential IP addresses. This capability allows operators to bypass geolocation restrictions and evade fraud detection systems while conducting account takeovers and transaction fraud. The malware uses commercial-grade obfuscation through Golden Encryption and establishes persistence through Accessibility Service abuse.
Cybercriminals are exploiting the recent Claude Code leak incident by using it as a social engineering tactic to deliver malware through GitHub repositories. The attackers have created trojanized versions of the leaked Claude source code, distributing malicious payloads including Vidar stealer version 18.7 and GhostSocks trojan. The campaign demonstrates rapid opportunistic exploitation of high-profile security incidents, with compromised repositories serving as delivery mechanisms. Organizations are advised to implement Zero Trust architecture to mitigate risks from shadow AI instances and trojanized Claude agents. Multiple GitHub repositories have been identified hosting the malicious code, with command and control infrastructure established across multiple IP addresses and domains.
On April 9, 2026, the cpuid.com website was compromised in a watering hole attack lasting approximately 19 hours. Download URLs for legitimate system administration tools CPU-Z, HWMonitor, HWMonitor Pro, and Perfmonitor 2 were replaced with links to malicious sites distributing trojanized versions. The malicious installers contained legitimate signed executables paired with DLL files named CRYPTBASE.dll that exploited DLL sideloading for C2 communication and payload delivery. Attackers reused infrastructure and code from a March 2026 fake FileZilla campaign, including the STX RAT as the final payload. Over 150 victims were identified globally, primarily individuals but including organizations in retail, manufacturing, consulting, telecommunications and agriculture sectors. The attack demonstrated poor operational security with reused indicators enabling rapid detection.
In February 2026, an attack chain was discovered that utilized a fraudulent Adobe Acrobat Reader download page to deceive victims into installing ConnectWise's ScreenConnect, a legitimate remote access tool exploited for malicious purposes. The attack employs sophisticated evasion techniques including heavy obfuscation, .NET reflection for in-memory payload execution, and dynamic code construction. A VBScript loader initiates the chain by downloading and executing obfuscated PowerShell commands that compile C# code entirely in memory. The loader manipulates the Process Environment Block to masquerade as legitimate Windows processes and abuses auto-elevated COM objects to bypass User Account Control without user prompts. This multi-layered approach successfully evades signature-based defenses and hinders forensic analysis while ultimately deploying ScreenConnect for unauthorized remote access.
A supply chain attack targeting the Axios npm package has been identified after threat actors compromised the npm account of the company's lead developer. Malicious versions (axios@1.14.1 and axios@0.30.4) were published containing a hidden dependency that executed postinstall scripts during npm installation. This automated execution downloaded and deployed a remote access trojan on affected systems without requiring user interaction, making it particularly dangerous for developer environments and CI/CD pipelines. The compromise resulted in full remote access capabilities, potential credential exposure including API keys and SSH keys, and possible insertion of malicious code into software builds. Detection platforms identified suspicious process execution chains involving npm spawning command interpreters and network utilities, followed by outbound connections to attacker-controlled infrastructure.
Storm-2755 is a financially motivated threat actor conducting payroll pirate attacks targeting Canadian employees. The campaign uses malvertising and SEO poisoning to redirect victims searching for terms like "Office 365" to fraudulent sign-in pages. By employing adversary-in-the-middle techniques, the attacker captures authentication tokens and session cookies, bypassing MFA protections. The actor maintains persistence by replaying stolen tokens and performs discovery to identify payroll and HR contacts. They impersonate compromised users to socially engineer HR staff or directly manipulate payroll systems such as Workday. Malicious inbox rules are used to hide attacker communications. These attacks have resulted in direct financial losses through redirected salary payments to attacker-controlled bank accounts.
