Kaspersky reported that the group known as Silver Fox is spreading the ValleyRAT backdoor, also called Winos 4.0, hidden inside a genuine but signed Chinese adware application called QN Wallpaper. By side-loading a malicious library through the trusted, signed program, the malware runs inside a process users are likely to have added to their antivirus exclusion lists, and it disables Windows Defender. Once active, it gives the operator full control, capturing keystrokes, clipboard contents, and screenshots and loading further modules. Kaspersky recorded more than 100,000 detections of ValleyRAT this year, mostly in China and India, and warns that adware and affiliate networks can be far more dangerous than they look.
A social-engineering campaign dubbed TerminalFix uses fake Cloudflare CAPTCHA pages, often served from compromised websites, to trick visitors into copying and running a malicious PowerShell command in their terminal. It is a refined take on the ClickFix technique, tuned to make complex scripts run more reliably, and it deploys a reverse-tunnel backdoor through a multi-stage chain involving DLL sideloading, hiding payloads inside images, and an outbound WebSocket connection for command and control. The campaign has hit organizations across several sectors. The core deception is simple to teach against: a legitimate CAPTCHA never asks you to paste and run commands in a terminal or Run dialog.
Researchers at Socket detailed a long-running campaign, active since early 2024, in which Chrome and Edge extensions delivered an extensible malware framework of sixteen modules to steal cryptocurrency, credentials, session tokens, and browsing data, and to inject ClickFix lures. Notably, several extensions started out legitimate and were only weaponized later, after their original developers handed over control and new owners pushed malicious automatic updates. One extension reached seventy thousand users before removal. The malware ran from the extension's background worker, opened an encrypted connection to its servers, and displayed fake wallet-recovery pages on real crypto sites to capture recovery phrases.
Researchers at Socket found a coordinated cluster of nineteen browser extensions, eighteen for Chrome and one for Edge, published over the past six months with code to steal cryptocurrency wallet secrets, drain funds, harvest credentials, and inject code into targeted websites. The extensions share code and tradecraft, suggesting a single campaign that may have run even longer. Because a browser extension can read and alter the pages a user visits, a malicious one that reaches a crypto user can quietly capture recovery phrases or redirect transactions. This continues a steady pattern of wallet-draining extensions slipping into official browser stores under the guise of useful tools.
Cosmos Labs warned that attackers are actively exploiting a flaw in its Cosmos EVM module, the component that gives Cosmos blockchains Ethereum compatibility, and urged affected chains to halt their validators. The bug lies in how the module handles state during nested transactions, letting an attacker manipulate balance and ownership tracking to move funds without authorization. Because many independent chains share the same module code, the weakness is systemic: separate chains were hit in the days before the ecosystem-wide warning, suggesting attackers generalized one exploit across the shared codebase. A patch existed from earlier in the year, but new exploitation shows the risk persists. It echoes supply-chain risk applied to blockchain infrastructure.
Researchers at VulnCheck found two undocumented factory implants in the firmware of routers made by the Chinese manufacturer ZBT, each giving a remote, unauthenticated attacker the ability to run commands as root. Named SPEAKINGSTONE and DARKLANTERN and tracked as CVE-2026-74232 and CVE-2026-74233, both scored around 9.3, require no privileges or interaction. SPEAKINGSTONE runs as a hidden service and beacons out over a fixed UDP port to a hardcoded command-and-control server; because it dials outward, it works from behind network address translation and normal egress filtering. The findings underscore the supply-chain risk of low-cost networking hardware with opaque firmware.
Researchers showed that Amazon Kiro, an AI-powered agentic development environment, can be turned against its user through prompt injection. Attacker-controlled content in a repository the developer opens can steer the Kiro agent into transmitting sensitive local information to an external server, abusing a feature called Kiro Powers that bundles model context protocol server configurations, steering files, and hooks. The developer only has to open the workspace and interact with the agent. It is part of a wider run of similar flaws in AI coding tools, where untrusted content or links quietly redirect an agent into exfiltrating data or executing code without any approval prompt. Updating the tool addresses the reported issue.
Academic researchers disclosed GPUThor, a Rowhammer attack that defeats the error-correcting memory that NVIDIA recommends as the defense against GPU Rowhammer. Demonstrated on Ampere-class workstation GPUs with GDDR6 memory, including the RTX A4000 through A6000 models common in AI and cloud infrastructure, it lets unprivileged code running on the GPU flip memory bits far more reliably than earlier attacks, finding an exploitable flip in about a minute. That enables denial of service, silent data corruption, and escalation to a root shell on the host. The researchers note error correction is not sufficient protection, which matters most where untrusted workloads share GPUs, as in multi-tenant cloud and AI platforms.
Researchers at OX Security found a campaign using two dozen npm packages as free phishing infrastructure rather than as malware aimed at developers. Each package is just a single HTML page, harmless to install, but once served through npm content-delivery mirrors like unpkg it becomes a live, fully rendered fake Cloudflare CAPTCHA page hosted on a trusted domain. The page then redirects victims to ClickFix-style phishing infrastructure, and while it currently forwards to a legitimate site, it can be reconfigured to deliver any phishing payload. The trick is not infecting people who install the packages, but abusing the registry and its mirrors as validated, reputable storage for attacker content.
A commercial phishing-as-a-service toolkit called Mirage2FA has hit around 4,500 organizations by abusing legitimate Microsoft 365 login flows to steal passwords and session cookies and bypass two-factor authentication. Because it captures the session cookie after a real login completes, the attacker inherits an authenticated Microsoft 365 session and any single-sign-on connected services, defeating multi-factor authentication. Researchers at ANY.RUN linked the campaign to more than 9,000 potential compromise events and found that nearly half of targeted addresses may have been affected, with most victims in the United States across technology, manufacturing, and education. Hijacking one session can expand into connected apps and internal workflows.