A compromised version of the Injective Labs TypeScript SDK npm package was published containing malicious code that exfiltrates cryptocurrency wallet private keys and mnemonic phrases. The malicious version 1.20.21 was published on June 8, 2026, through a compromised developer account with established repository access. The malware hooks key generation functions to capture sensitive wallet data and exfiltrates it via base64-encoded POST requests to legitimate Injective infrastructure endpoints, disguising the traffic. The threat actor amplified impact by publishing 17 additional scoped packages pinned to the malicious version. Though quickly detected and contained within hours, the compromised package received approximately 310 downloads. The package has roughly 50,000 weekly downloads and 87 dependent packages, presenting significant supply chain risk to cryptocurrency wallet implementations.
Vidar, a Malware-as-a-Service infostealer first identified in 2018, continues to be distributed through phishing campaigns targeting Korea in the first half of 2026. The threat actor uses phishing emails disguised as job applications and copyright infringement notices, with attachments appearing as Word documents but actually being executables. Vidar employs a Go-based packer, uses Dead Drop Resolver technique via Telegram and Steam profiles to obtain C&C addresses, and implements anti-debugging and anti-VM techniques. The infostealer exfiltrates sensitive information including browser credentials, cookies, browsing history, cryptocurrency wallet data, Discord tokens, Telegram information, Steam data, Azure credentials, and screenshots. Configuration information is downloaded in JSON format, and data collection is performed based on received flags and additional downloaded conditions.
Salat Stealer is a Go-based information stealer that performs deep system reconnaissance and extracts sensitive data from compromised hosts. It targets browser credentials, cryptocurrency wallets, and communication platforms like Discord and Steam. The malware features advanced surveillance capabilities including desktop streaming, audio/video capture through microphone and webcam, and local file exfiltration. A notable distribution campaign bundled Salat Stealer with Xeno Executor, a gaming utility tool, transforming it into a full compromise vector. The malware employs sophisticated evasion techniques including disabling Windows Defender features through multiple PowerShell scripts, establishing persistence via registry run keys, and using token impersonation of lsass.exe to obtain elevated privileges. Loaders written in batch script and Rust programming language obfuscate deployment and bypass security controls.
A new macOS ClickFix campaign employs fake CAPTCHA pages to deceive users into executing malicious Terminal commands. The attack chain downloads and invisibly mounts a DMG file containing a self-signed information-stealer application bundle. This payload, assessed as belonging to the AMOS (Atomic macOS Stealer) lineage—specifically the Odyssey variant—prompts users for passwords through fake System Preferences dialogs. The stealer harvests extensive data including browser credentials, cryptocurrency wallet information from 13 standalone applications and 201 browser extensions, messaging app data, Apple Notes, Safari cookies, and macOS keychain entries. Exfiltrated data is compressed and sent to two command-and-control servers. The malware establishes persistence via LaunchAgent and trojanizes legitimate cryptocurrency applications including Ledger Live and Trezor Suite, replacing them with compromised versions downloaded from attacker infrastructure.
In May 2026, security researchers observed an attack campaign distributing KimJongRAT through GitHub and other legitimate services. KimJongRAT, used by the North Korean APT group Kimsuky since 2013, combines information stealing and remote access capabilities. The infection chain begins with phishing emails containing shortened URLs redirecting to GitHub Releases hosting malicious ZIP files. Victims execute LNK files that download HTA files from GitHub, which then retrieve subsequent payloads from Google Drive. Recent variants demonstrate significant evolution: they now dynamically fetch C2 addresses from external sources rather than hardcoding them, enabling operators to maintain persistent access despite infrastructure takedowns. Additionally, new versions include MeshAgent RMM installation for redundant access. The campaign exemplifies Living Off Trusted Sites (LOTS) techniques, abusing legitimate platforms like GitHub, Google Drive, and Dropbox to evade detection.
ESET Research contributed to a global disruption operation targeting the Amadey botnet and Stealc infostealer, both malware-as-a-service offerings. The operation, coordinated by Microsoft Digital Crimes Unit, BitSight, Lumen, and MBSD, impacted approximately 50 domains and nearly 200 active IP-based command and control servers. ESET provided technical analyses, statistical information, C&C server lists, encryption keys, campaign identifiers, and affiliate-level insights gathered from three years of tracking. Both malware families operate through affiliate networks where operators deploy their own infrastructure, making disruption efforts particularly challenging. Amadey primarily functions as a modular loader distributing additional payloads, while Stealc focuses on credential theft from browsers, crypto wallets, and applications. The largest Amadey botnet cluster accounted for 34% of all samples and distributed an average of 14 payloads per victim, operating a pay-per-install model that monetized compromi...
Infostealers remain among the most pervasive cybercrime threats, silently harvesting passwords, cookies, and session tokens that enable enterprise breaches. StealC is a malware-as-a-service infostealer written in C++ that collects credentials from browsers, cryptocurrency wallets, messaging applications, email clients, and gaming platforms while functioning as a secondary loader. Amadey operates as a modular backdoor loader active since 2018, delivering downstream payloads including StealC, Lumma Stealer, and ransomware through various backdoor commands. Both operate on commodity rental models where stolen credentials flow through underground markets to access brokers who resell enterprise access. On June 24, 2026, Microsoft's Digital Crimes Unit coordinated with Europol to disrupt over 200 malicious command-and-control domains supporting these operations, using AI-assisted analysis tools including Microsoft Copilot for binary analysis and configuration extraction.
Between February and May 2026, researchers identified five malicious skills on ClawHub, OpenClaw's AI agent marketplace, that evaded detection by VirusTotal and ClawScan. The threats included two macOS infostealers communicating with command-and-control infrastructure, one skill using file padding to bypass scanner thresholds, and two novel agentic threats exploiting the AI supply chain for financial gain. The infostealers delivered payloads including AMOS malware through Base64-encoded droppers and paste-site redirects. One skill implemented runtime affiliate injection by forcing agents to recommend products through malicious referral links, while another orchestrated a front-running scheme using coordinated AI agents to manipulate cryptocurrency token launches. These attacks demonstrate how malicious actors exploit semantic instruction hijacking and the lack of isolation between skill logic and agent authority to compromise AI agent ecosystems.
A sophisticated phishing campaign was identified distributing multiple malware families through a multi-stage loader utilizing steganography and fileless techniques. The infection chain begins with archive attachments containing files disguised as financial documents, primarily targeting Indian organizations using names related to GST, NEFT, RTGS, and IMPS transactions. The loader employs in-memory execution to avoid disk-based artifacts and uses embedded .NET Bitmap objects to conceal payloads. Various malware families have been deployed including Remcos RAT, Agent Tesla, MassLogger, Phantom Stealer, Dark Cloud, Red Line Stealer, Snake keyloggers, Formbook, and xworm. The final payloads establish persistence through registry Run keys, perform process hollowing, steal browser credentials, record audio and webcam, and exfiltrate data to command-and-control infrastructure. The campaign exhibits characteristics of a loader-as-a-service operation serving multiple threat actors globally.
A previously undocumented Windows loader designated as OXLOADER delivers the CASTLESTEALER infostealer through malicious Google Ads campaigns, achieving remarkably low detection rates. The loader employs multiple obfuscation layers including control-flow flattening, opaque predicates, and mixed Boolean-Arithmetic techniques, along with self-modifying decryption stubs and abuse of the Windows .reloc section for shellcode staging. Distribution occurs via malvertising impersonating Node.js installations, redirecting victims through intermediary domains to Storj-hosted batch scripts. The loader implements five anti-VM and language checks, including CIS-region and Russian-language exclusions, suggesting a financially motivated Russian-speaking threat actor. OXLOADER uses DonutLoader to deliver the .NET-based CASTLESTEALER payload in memory, evading traditional detection mechanisms through deliberate engineering choices.
This analysis covers infostealer distribution trends observed during May 2026, based on automated collection systems and diagnostic logs. Distribution occurred primarily through illegal software disguised as cracks and keygens, as well as email campaigns. ACRStealer, Remus, and LummaC2 were most prevalent, with distribution via domains including Mediafire and AWS S3 buckets. Microsoft was the most impersonated company, followed by Auslogics and NVIDIA. EXE files represented 78.9% of execution types, while DLL side-loading accounted for 21.1%. macOS environments saw ClickFix techniques and malicious Bash scripts, with 142 scripts and 12 C2 domains identified. Email campaigns distributed AgentTesla and DarkCloud. Remus showed significant growth, comprising 36% of distributions. LummaC2 remained the most prevalent overall variant.
More than 140 Mastra npm packages were compromised through a supply chain attack that injected a typosquatted dependency called easy-day-js. A single npm account published malicious versions within a short timeframe, affecting packages including @mastra/core with over 918K weekly downloads. The attack executes during npm install via a postinstall hook, deploying a two-stage payload. The first stage disables TLS validation and downloads a second-stage implant that installs cross-platform persistence on Windows, macOS, and Linux. This implant functions as a command-and-control client that steals cryptocurrency wallet inventories from 166+ browser extensions, harvests browser history, and can execute arbitrary code sent by operators. The malicious code executes before developers import packages, compromising systems during installation.
Since late 2025, cybercriminals have been exploiting Wallpaper Engine, a popular live wallpaper application on Steam, to distribute malware through Steam Workshop. Attackers target primarily Chinese and Russian gamers by embedding malicious code within application wallpapers shared on the platform. These compromised wallpapers deliver various malware types including infostealers, backdoors, crypto miners, and ransomware. One analyzed sample dropped DarkKomet backdoor while hijacking Steam sessions to steal account credentials. The malware modifies system libraries to locate Steam installations and exfiltrate data to attacker-controlled servers. Compromised accounts are then used to upload additional malicious wallpapers. The diverse malware families suggest multiple independent hacking groups are exploiting this distribution method. Infected wallpapers received thousands of downloads before removal, with 89% of infections occurring in China.
Threat actors are leveraging TikTok and Instagram Reels to distribute the Vidar infostealer through fake software tutorials. Two distinct campaigns use short-form videos disguised as tutorials for unlocking premium software like Spotify. The first campaign uses accounts mimicking official Windows profiles with AI-voiced clips instructing users to run PowerShell commands that download Vidar from lookalike domains. One video achieved over 100,000 views. The second campaign uses ordinary accounts posting music-backed clips that bait users in comments to receive malicious links via direct message. These campaigns exploit platform recommendation algorithms by encouraging saves and shares. Vidar is sold as a service for $300 lifetime license and harvests credentials, financial data and authentication tokens.
Threat actors are exploiting short-form video platforms like TikTok and Instagram Reels to conduct social engineering attacks. Two distinct campaign methods have been identified: professional-looking fake tutorials with AI-generated voiceovers promising free premium software, and casual videos showcasing premium features to generate engagement through comments. Both approaches direct victims to malicious websites hosting infostealer malware, particularly Vidarstealer. The campaigns leverage platform algorithms through high engagement rates including saves, shares, and comments. Attackers use multiple accounts with Windows-themed branding and manipulate PowerShell commands to download malicious executables. These techniques are difficult to counter as creators can delete warning comments and platform reporting mechanisms prove ineffective. The attacks target non-technical users seeking free access to premium services like Spotify, Microsoft Office, and other software, making social media feeds an emerging p...
This analysis examines Gamaredon's (UAC-0010, Armagedon) advanced espionage operations targeting Ukrainian government, military, and critical infrastructure. The FSB-operated group deploys GammaSteel, a sophisticated stealer operating almost entirely from memory using Windows DPAPI encryption and storing 71 distinct payload functions in the HKCU\Printers registry key. The malware employs three concurrent data acquisition mechanisms: timed drive scans, USB monitoring for air-gapped systems, and real-time file surveillance. Exfiltration occurs via legitimate S3-compatible cloud storage (Tebi.io) with fallback to operator-controlled servers. The infection chain extensively uses VBScript for evasion, Dead Drop Resolvers on platforms like Telegram and Mastodon for C2 configuration, and includes bidirectional backdoor capabilities enabling arbitrary remote code execution. Infrastructure demonstrates high automation with servers rotated approximately every 24 hours.
Threat actors exploited the EtherHiding technique to store ClearFake payload routing instructions within smart contracts on the BNB Smart Chain testnet, creating an immutable command-and-control infrastructure that cannot be taken down. The attack began with injected JavaScript on a compromised Swiss website that queried blockchain contracts to deliver malicious payloads. Victims passing anti-analysis checks were fingerprinted by operating system and routed to platform-specific ClickFix social engineering overlays. The campaign simultaneously deployed SectopRAT, a .NET-based remote access trojan capable of browser session hijacking, and ACRStealer, a C++ infostealer targeting credentials and cryptocurrency wallets. An on-chain execution tracker confirmed each compromise in real time. Four smart contracts shared a single deployer wallet, with the oldest deployed nearly a year before analysis, indicating a long-running, actively maintained operation.
Attackers are concealing .NET infostealers within seemingly innocuous images to evade detection. A phishing campaign uses TXZ archive attachments with invoice-themed lures to initiate infection. The embedded JavaScript leverages environment variables to hide malicious commands, launching PowerShell to decode and decrypt payloads. PawsRunner, a steganography loader, extracts encrypted data from PNG images containing cat photos. This loader evolved from simple PE downloads to sophisticated steganographic extraction with fallback mechanisms. The final payload, PureLogs version 5.0.0, is a comprehensive infostealer from the Pure family that harvests credentials from browsers, cryptocurrency wallets, password managers, communication apps, and other applications. It employs extensive async/await patterns and communicates with command and control infrastructure via HTTPS using multiple endpoints to exfiltrate encrypted and compressed stolen data.
Financially motivated eCrime actors are conducting an ongoing infostealer campaign targeting software developers through SEO poisoning techniques. The operation impersonates AI platforms including Gemini CLI and Claude Code, as well as developer tools like Node.js, Chocolatey, and KeePassXC. Attackers position fake domains above legitimate search results, directing victims to malicious installation pages that deliver fileless PowerShell-based infostealer malware. The malware executes entirely in memory, disables Windows Defender telemetry by patching ETW and AMSI, and harvests credentials from browsers, collaboration platforms, VPN clients, and cloud storage. Stolen data includes OAuth tokens, CI/CD credentials, and corporate VPN details, providing direct enterprise network access. The campaign leverages bulletproof hosting infrastructure and over 30 typosquatted domains registered between March and April 2026, primarily targeting users in the United States and United Kingdom.
A sophisticated infostealer operation was discovered masquerading as a cryptocurrency trading application called Tralert FX. The malicious MSI installer achieved only 3/52 AV detections by using a valid EV code signing certificate from a likely front company, AgilusTech LLC. The campaign has been active since June 2025, utilizing a three-module malware kit that includes system reconnaissance, keylogging, and browser credential theft capabilities. Stolen data is exfiltrated through five GitLab repositories via automated git commits on 30-minute cycles. Hardcoded credentials exposed the entire backend infrastructure, revealing over 4,100 commits, 90+ compromised hosts, and ongoing victim compromise. The operation demonstrates clear financial motivation with focus on cryptocurrency traders for account takeover. Three ProtonMail-linked GitLab accounts operate the infrastructure, assessed as a single operator or small team. The final payload is MoonPeak, a custom variant of XenoRAT.
