All news with #malware tag

🔬 Netskope Threat Labs tested whether large language models can generate operational malware by asking GPT-3.5-Turbo, GPT-4 and GPT-5 to produce Python for process injection, AV/EDR termination and virtualization detection. GPT-3.5-Turbo produced malicious code quickly, while GPT-4 initially refused but could be coaxed with role-based prompts. Generated scripts ran reliably on physical hosts, had moderate success in VMware, and performed poorly in AWS Workspaces VDI; GPT-5 raised success rates substantially but also returned safer alternatives because of stronger safeguards. Researchers conclude LLMs can create useful attack code but still struggle with reliable evasion and cloud adaptation, so full automation of malware remains infeasible today.

⚠️ The second wave of the Shai-Hulud supply-chain attack has moved from npm into the Maven ecosystem after researchers found org.mvnpm:posthog-node:4.18.1 embedding the same setup_bun.js loader and bun_environment.js payload. The artifact was rebundled via an automated mvnpm process and was not published by PostHog; mirrored copies were purged from Maven Central on Nov 25, 2025. The campaign steals API keys, cloud credentials and npm/GitHub tokens by backdooring developer environments and injecting malicious GitHub workflows, affecting thousands of repositories.

🛡️ Researchers at Huntress uncovered a ClickFix campaign that hides malware inside the RGB pixels of PNG images on a fake Windows Update page, tricking victims into pasting and running commands. The delivered payloads include the LummaC2 infostealer and the Rhadamanthys malware family, with active domains observed after a mid-November takedown. Huntress warns the steganographic technique and the realistic Windows Update motif increase the attack's stealth, and recommends disabling the Windows Run dialog and strengthening endpoint monitoring.

🛡️ A malicious Chrome extension named Crypto Copilot was found injecting covert Solana transfers into Raydium swap transactions, diverting funds to an attacker-controlled wallet. Published by "sjclark76" on May 7, 2024, the add-on remains available on the Chrome Web Store with 12 installs. The extension appends a hidden SystemProgram.transfer to each swap before signature, charging a minimum of 0.0013 SOL (and applying a 2.6 SOL/0.05% rule) while obfuscating its code to evade detection. It also contacts backend domains to register wallets and report activity, giving a false veneer of legitimacy.

🛡️Huntress warns of an evolved ClickFix campaign that uses a convincing full‑screen Windows Update splash and steganographic PNGs to trick employees into pasting and running commands. Those commands deliver loaders that in turn deploy LummaC2 and Rhadamanthys infostealers. The firm reports a 313% increase in ClickFix incidents over six months and noted multiple active lure domains even after the Nov 13 Operation Endgame takedown. Primary mitigation advice is to disable the Windows Run dialog via Registry or GPO and pair user awareness with endpoint monitoring and EDR.

🔒 AWS has launched Network Firewall Proxy in public preview, providing centralized controls to block data exfiltration and malware injection across application traffic. In explicit proxy mode you can set up filters in just a few clicks to control outbound requests and the responses your applications receive, protect against domain or SNI spoofing, and restrict access to trusted domains or IPs. The service supports TLS inspection and granular HTTP header filtering, and emits detailed logs to Amazon S3 and AWS CloudWatch. Preview access is free in US East (Ohio).

⚠️ Cybersecurity researchers report a JackFix campaign that uses fake Windows Update pop-ups on cloned adult sites to trick users into running mshta.exe and PowerShell commands. According to Acronis and Huntress, the attack chain leverages obfuscation, privilege escalation and can deploy multiple stealers including Rhadamanthys, RedLine and Vidar. Organizations are advised to train users and consider disabling the Windows Run box via Group Policy or Registry changes to reduce risk.

🦊 Jamf Threat Labs reports a macOS malware chain, named FlexibleFerret, that employs staged scripts, credential‑harvesting decoys and a persistent Go-based backdoor to maintain long-term access. The campaign uses a second-stage shell script that reconstructs download paths and fetches different payloads for arm64 and Intel systems, then unpacks and runs a loader while writing a LaunchAgent for persistence. A decoy app mimics Chrome permission prompts and a Chrome-style password window to steal credentials, which are exfiltrated via the legitimate Dropbox API. The final stage invokes a Golang backdoor, CDrivers, that provides remote command-and-control and extensive data-theft capabilities.

🛡️ Cybersecurity researchers disclosed a campaign that leverages Blender .blend files hosted on public asset sites to deliver the information stealer StealC V2. Malicious .blend assets contain embedded Python scripts that execute when Blender's Auto Run is enabled, fetching PowerShell code and two ZIP archives — one deploying StealC V2 and the other a secondary Python stealer. Vendors advise keeping Auto Run disabled and verifying asset sources.

🛡️ Unit 42 examines the growing threat of malicious large language models that have been intentionally stripped of safety controls and repackaged for criminal use. These tools — exemplified by WormGPT and KawaiiGPT — generate persuasive phishing, credential-harvesting lures, polymorphic malware scaffolding, and end-to-end extortion workflows. Their distribution ranges from paid subscriptions and source-code sales to free GitHub deployments and Telegram promotion. The report urges stronger alignment, regulation, and defensive resilience and offers Unit 42 incident response and AI assessment services.

⚠️ Researchers at Morphisec observed a Russian-linked campaign using malicious Blender .blend files uploaded to 3D model marketplaces to deliver the StealC V2 infostealer. The embedded Python in the .blend fetches a loader from a Cloudflare Workers domain, which runs a PowerShell script to download two ZIP archives, unpack them into %TEMP%, drop LNK shortcuts into the Startup folder for persistence, and deploy both the StealC payload and an auxiliary Python stealer. Users are advised to disable Blender's Auto Run for Python scripts and treat downloaded 3D assets like executables, testing unknown files in sandboxed environments.

🔒 Researchers warn of ClickFix attack variants that display a realistic full‑screen fake Windows Update animation in the browser to trick users into pasting commands that execute malware. Operators use steganography to hide AES‑encrypted shellcode inside PNG pixel data and leverage mshta, PowerShell, and a .NET Stego Loader to reconstruct and run payloads. Huntress observed delivery of LummaC2 and Rhadamanthys info stealers and a dynamic evasion ctrampoline technique to hinder analysis. A law enforcement takedown in November disrupted payload delivery on some fake update domains.

⚠️ Superbox media streaming boxes sold through retailers like BestBuy and Walmart have been found running intrusive, unofficial apps that can enlist buyers' Internet connections into distributed residential proxy networks and botnets. Censys researchers observed devices phoning home to Tencent QQ and a proxy service called Grass IO, and installing tools such as tcpdump and netcat while performing DNS hijacking and ARP spoofing. The boxes require removing Google Play and installing a third-party app store, increasing the risk of unauthorized relays, advertising fraud, and account takeovers. Consumers are advised to avoid uncertified Android TV devices and follow FBI and EFF guidance on suspicious app marketplaces.

🛠️ Morphisec researchers have uncovered a six-month campaign embedding StealC V2 inside weaponized Blender .blend files distributed via marketplaces such as CGTrader. When opened with Blender's Auto Run enabled, concealed Python scripts fetch loaders from workers.dev domains and initiate a multistage infection that deploys PowerShell components and Python-based stealers. The malware establishes persistence with LNK files and communicates with Pyramid-linked C2 servers to retrieve encrypted payloads. Morphisec says its deception-based protection thwarts credential theft by injecting decoy credentials and terminating processes before exfiltration.

🔍 Check Point's Harmony Mobile Detection Team discovered a broad Android adware campaign on Google Play that operated as a persistent background advertising engine. Masquerading as benign utilities and emoji editors, the apps continued running after closure or reboot, quietly consuming battery and mobile data. The campaign, dubbed GhostAd, comprised at least 15 related apps, with five still available at discovery.

🔔 Matrix Push C2 is a browser-native, fileless C2 platform that leverages web push notifications, fake alerts, and link redirects to distribute phishing links across operating systems. Attackers social-engineer users into allowing notifications on malicious or compromised sites, then send branded, OS-like alerts with action buttons that redirect victims to fraudulent landing pages. Sold as a MaaS kit via Telegram and cybercrime forums, it includes a web dashboard, analytics, URL shortening, configurable templates (e.g., MetaMask, Netflix, PayPal), and tiered crypto-paid subscriptions.

⚠️ Kaspersky describes a campaign in which attackers used the AI-powered web builder Lovable to mass-generate convincing fake vendor pages that host malicious installers. Those pages distribute a custom, attacker-signed build of the legitimate remote administration tool Syncro, which installs silently and grants full remote access. Because the payload is a legitimate admin tool altered for abuse, detection is difficult and victims risk data theft and loss of cryptocurrency funds.

🔔 BlackFrog has identified a new command-and-control platform, Matrix Push C2, that abuses browser push notifications to deliver phishing and malware. The campaign social-engineers users into allowing notifications and then issues realistic system-style alerts that redirect victims to malicious sites. Described as fileless, the technique leverages the browser notification channel rather than an initial executable. The platform includes a web dashboard with real-time client visibility, analytics and templates impersonating services like MetaMask, Netflix and PayPal.

🛡️ Kaspersky researcher Lisandro Ubiedo details an expanding Windows-focused botnet named Tsundere that retrieves and executes arbitrary JavaScript from remote command-and-control servers. The threat, active since mid‑2025, has been distributed via fake MSI installers and PowerShell scripts that deploy Node.js, install dependencies (ws, ethers, and pm2) and establish persistence. Operators fetch WebSocket C2 addresses from an Ethereum smart contract to rotate infrastructure, while a control panel enables artifact building, bot management, proxying, and an on-platform marketplace.

🔍 Group-IB researchers reveal a multi-year ATM fraud campaign by UNC2891 that targeted two Indonesian banks and extended well beyond a Raspberry Pi infiltration. The campaign combined sophisticated malware — including the CAKETAP rootkit — with an extensive money-mule operation that recruited via Google ads and Telegram. Cloned card equipment was shipped to mules, who withdrew cash with real-time TeamViewer support or phone coordination. Group-IB warns banks to reassess ATM security and monitoring.