Intelligence Feed

1mo ago · talos

In the first half of 2026, ransomware incidents in Japan increased slightly by 4.7%, with The Gentlemen emerging as the most active group, responsible for 14 incidents. The group operates via a Ransomware-as-a-Service (RaaS) model and uses a double-extortion tactic, leveraging infrastructure including AdaptixC2 for command-and-control. Evidence from Russian-language artifacts in scripts and bash history suggests Russian-speaking actors are involved. Qilin, the second most active group, showed signs of using generative AI in developing attack scripts, with code exhibiting structured, LLM-like patterns in tools for deploying ransomware and wiping backups.

1mo ago · hacker-news

A new JavaScript stealer named WeaselBiscuit has been distributed through 13 malicious npm packages, targeting developers by harvesting Chrome extension storage data. The malware is lightweight and memory-resident, importing and executing its payload from an Npoint.io dead drop. It communicates with a C2 server at 103.170.217.184:8787 to receive commands and exfiltrate data, including clipboard contents and keystrokes on Windows. While it shares functional and tradecraft similarities with DPRK-linked malware BeaverTail and OtterCookie, no definitive attribution has been made.

1mo ago · hacker-news

A vulnerability dubbed Plugin4Shell affects four AI coding agents—Anthropic Claude Code, OpenAI Codex, GitHub Copilot, and Google Gemini CLI—allowing repository owners to swap pinned plugin code with malicious versions by exploiting how commit hashes are validated. The flaw arises because agents fetch a specific commit hash but fail to verify that the retrieved code matches it, enabling attackers to create a branch name that mimics the hash and point it to different, malicious code. While Anthropic and OpenAI have released patches in versions 2.1.179 and 0.146.0 respectively, GitHub Copilot has no fix available and Google will not patch the retiring Gemini CLI. The attack can lead to unauthorized access to user files, credentials, and systems, especially when combined with background auto-update features enabled by default in some agents.

1mo ago · hacker-news

Transparent Tribe (APT36), a Pakistan-aligned threat actor, has launched a new campaign dubbed Operation RapidRust, targeting government and defense entities in India and Afghanistan. The group deployed a Rust-based backdoor called RUSTYSHADE that uses private GitHub repositories for encrypted command-and-control (C2) communications. Additional tools include RUSTYMOVE, a USB propagation tool, and file stealers PSNATCH (PowerShell) and BASHNATCH (bash) for Windows and Linux systems. Attackers used typosquatted domains impersonating Indian news sites to host malicious payloads and conducted post-compromise activities including reconnaissance and lateral movement.

1mo ago · hacker-news

WordPress patched a vulnerability dubbed Click2Shell, which allows a specially crafted URL to force the installation of a theme from the WordPress.org directory when clicked by a logged-in administrator, without requiring additional interaction. While the core flaw alone only installs a legitimate theme, it can be chained with a vulnerability in the installed theme to achieve remote code execution. The attack exploits inconsistent parsing of URL parameters between WordPress.org and the administrator's browser, leading to unintended automatic clicks on the Install button. The full exploit chain was demonstrated using the 'Mobile Repair Zone' theme, which contains a handler that downloads and executes arbitrary code without authentication.

1mo ago · bleeping-computer

A malware campaign is distributing a new information stealer named Rapuncel through SEO-optimized fake GitHub repositories impersonating LastPass and 39 other software brands. The attack delivers the Rapuncel infostealer alongside a Microsoft-signed kernel driver, Alinubx.sys, which disables 145 antivirus and EDR products by exploiting kernel-level access. The malware steals credentials from browsers, cryptocurrency wallets, Discord, Steam, Telegram, Windows Credential Manager, and specific documents, while also capturing screenshots and system information, exfiltrating data to a C2 server via raw TCP.

1mo ago · bleeping-computer

Gyazo, a cloud-based screenshot and screen-recording platform, suffered a data breach on September 11, 2026, when attackers exploited a server vulnerability to access its database and steal approximately 23.62 million user records. The compromised data includes personal information such as names, email addresses, password hashes, session IDs, and integration tokens, as well as metadata for 490 million images, some of which may have been private. Gyazo temporarily suspended the service to investigate and mitigate the breach, urging users to change their passwords and remain vigilant against potential phishing or follow-up attacks.

1mo ago · hacker-news

Microsoft has patched a critical vulnerability, CVE-2026-85889, in Azure AI Foundry with a CVSS score of 10.0, stemming from missing authentication for a critical function that could allow unauthorized attackers to escalate privileges over the network. The flaw has been mitigated by Microsoft with no customer action required, and there is no evidence of exploitation in the wild. Additional critical vulnerabilities were also patched across Microsoft 365 Copilot, Azure Database for PostgreSQL, Azure Cosmos DB, and Windows components, including locally exploitable privilege escalation flaws.

1mo ago · hacker-news

The Iran-linked threat actor known as Handala Hack, attributed to the MOIS-affiliated group Void Manticore, has been linked to a multi-stage malware campaign using the Telegram-based backdoor HEAVYGRAM and the Delphi-based utility CRUDEEXCLUDE. HEAVYGRAM is a Python-based backdoor that uses Telegram for command-and-control, enabling remote command execution, data exfiltration (including Telegram and WhatsApp data and saved passwords), screenshot capture, and persistence via Windows Registry keys. The malware is delivered through social engineering on messaging platforms, often disguised as legitimate applications like Pictory, KeePass, or Telegram, and leverages PowerShell and DLL sideloading for execution and evasion. CRUDEEXCLUDE prepares the environment by configuring Microsoft Defender exclusions and staging HEAVYGRAM and other payloads.

1mo ago · hacker-news

A critical vulnerability in Docker Sandboxes for macOS, tracked as CVE-2026-77179, allows malicious guest code running inside a virtual machine to escape the sandbox and read or modify arbitrary files on the host system by exploiting symlink handling in the virtio-fs host server. The flaw affects Docker Sandboxes versions 0.28.0 to 0.41.9 and was patched in version 0.42.0. A second high-severity issue, CVE-2026-79994, enables symlink-based path traversal to access Unix domain sockets outside the authorized workspace. No active exploitation has been reported, and both vulnerabilities were addressed in the same update.

1mo ago · hacker-news

A threat actor known as CL-CRI-1171 operated a pay-per-install (PPI) marketplace using YouTube and SEO poisoning to distribute malware, delivering payloads such as Docro Hijacker, ARKTunnel, and the Insomnia RAT via the OfferLoader custom loader. Separately, unauthenticated LocalAI instances were exploited to gain command execution and exfiltrate sensitive data, including military and AWS credentials. A critical VMware vCenter vulnerability, CVE-2026-59310, is being actively exploited by ransomware gangs for remote code execution. Additionally, Cyclops Blink has reemerged in a new variant targeting Cisco FMC devices, leveraging CVE-2026-20079 and CVE-2026-20316 for network reconnaissance and persistence.

1mo ago · hacker-news

A critical vulnerability, CVE-2026-91843, in Check Point's Security Management and Log Servers allows unauthenticated remote attackers to execute code as root via a stack overflow in the login process when a long username is provided. The vulnerability affects multiple versions of Check Point's R80, R81, and R82 branches, including R82.20, and impacts standalone, Log Server, and Multi-Domain deployments. Exploitation is possible over the network through the Trusted Clients setting, which must be restricted to prevent unauthorized access. Check Point has released a LivePatch fix, but no known in-the-wild exploitation has been reported as of September 17, 2026.

1mo ago · hacker-news

RatHat is a newly identified Android malware attributed to China-based threat actors that leverages AI to interact with compromised devices. It is distributed via smishing and malvertising campaigns leading to third-party download portals, where users are tricked into installing malicious APKs. The malware abuses Android Debug Bridge (ADB) to gain shell-level privileges and maintain persistence even after uninstallation by using a reverse tunnel via an FRP client to a C2 server. It can harvest credentials, record screens, intercept SMS, and log keystrokes, while using AI to dynamically navigate the device UI based on Accessibility tree data.

1mo ago · hacker-news

A threat actor posing as a bug bounty hunter has been linked to the development and distribution of PhantomRaven, a JavaScript-based information stealer targeting developers via malicious npm packages. The malware, likely generated using a large language model, retrieves remote dependencies to evade detection and steals authentication tokens, CI/CD secrets, and system information. The actor has used compromised environments to claim bug bounties from multiple organizations and has also attempted to distribute similar malware through PyPI.

1mo ago · bleeping-computer

A new Android malware named RatHat has been identified by Zimperium zLabs, which leverages AI to automate remote device navigation and control. The malware is distributed via malvertising, SMS, and phishing sites that prompt users to download malicious APKs outside Google Play. It abuses Android Accessibility permissions and enables Developer Options and Wireless Debugging to establish persistent ADB-level access through a Go-based agent. The malware includes credential harvesting, SMS interception, and lock-screen bypass capabilities, and uses AI to dynamically interpret the device interface for real-time remote control. Anti-analysis techniques and persistence mechanisms make it resilient to removal and detection.

1mo ago · bleeping-computer

Check Point Software disclosed a critical vulnerability, CVE-2026-91843, that allows unauthenticated attackers to achieve remote code execution with root privileges on Security Management Server and Log Server instances due to a stack-based buffer overflow in the login process. The vulnerability is exploitable without user interaction and affects all configurations, regardless of VPN usage. While not currently exploited in the wild, Check Point has provided detection guidance via specific log alerts and recommends immediate patching or implementation of mitigations such as restricting access to trusted IPs. The company recently patched other critical flaws, including CVE-2026-85103 and CVE-2026-85102, which are also under active exploitation threat.

1mo ago · socket-dev

The PolinRider campaign continues to spread through compromised GitHub accounts and Packagist, leveraging Git-based infrastructure to inject malicious code into development versions of popular packages. Researchers identified malicious activity in the dev branches of the visanduma/nova-two-factor Packagist package, which has over 700,000 downloads. The attackers use compromised developer accounts to insert obfuscated JavaScript into configuration files and PHP entry points, enabling automatic execution upon repository access or build processes. The campaign employs staged payload delivery via dead-drop resolvers and maintains persistence by rewriting Git history and exploiting IDE integrations such as VS Code tasks.

1mo ago · bleeping-computer

Brevo suffered a supply-chain attack where attackers stole a long-lived Cloudflare API key hardcoded in source code, allowing them to deploy a malicious Cloudflare Worker that injected malicious ClickFix scripts into Brevo's web assets. The compromised scripts were distributed to customer sites embedding Brevo components, affecting up to 100,000 websites. On WordPress sites, the attack attempted to upload a malicious plugin called 'Web Media Optimizer' that acts as a persistent backdoor and injects further malicious JavaScript, including fake Cloudflare verification pages prompting users to run harmful commands.

1mo ago · hacker-news

The U.S. Department of Justice, in coordination with the FBI and the Royal Canadian Mounted Police, seized domains associated with NightmareStresser, a DDoS-for-hire service that has been used to launch hundreds of thousands of attacks since 2022. The service, advertised as a stress-testing tool, enabled users to conduct Layer 4 and Layer 7 DDoS attacks with features like 'Stop All' flood controls and cryptocurrency payments. The takedown is part of Operation PowerOFF, a broader law enforcement initiative targeting criminal DDoS infrastructure globally.

1mo ago · hacker-news

Cisco has disclosed a critical zero-day vulnerability, CVE-2026-76460, affecting its Identity Services Engine (ISE) and ISE Passive Identity Connector (ISE-PIC) that is actively being exploited. The flaw, rated CVSS 10.0, allows unauthenticated remote attackers to bypass authentication by sending a crafted request to an affected API endpoint, potentially leading to full system compromise with root privileges. Cisco confirms active exploitation and advises immediate patching, as no workarounds exist. The U.S. CISA has added the vulnerability to its Known Exploited Vulnerabilities catalog, mandating federal agencies to patch by September 19, 2026.

1mo ago · hacker-news

A breach at Gyazo, an image-sharing service operated by Helpfeel, exposed approximately 23.62 million user records and 490 million image metadata records. The attacker exploited a vulnerability in Gyazo's image upload server to gain unauthorized access, execute arbitrary commands, and extract sensitive data including email addresses, password hashes, session IDs, and image metadata such as image IDs, upload IP addresses, User-Agent strings, EXIF location data, OCR-extracted text, and hashed passphrases for private images. Helpfeel confirmed the breach on September 14, 2026, reported it to Japanese authorities, and took steps to block the attacker and fix the vulnerability, though it did not disclose the specific nature of the flaw. The company temporarily disabled access to some images to prevent further unauthorized viewing and urged all users to change their passwords due to the risk of credential reuse.

1mo ago · hacker-news

The Internet Systems Consortium (ISC) released updates for BIND 9, addressing 14 security vulnerabilities, including a high-severity flaw allowing an unauthenticated attacker to crash a DNS-over-HTTPS (DoH) server with a single malformed request. Several other flaws can lead to denial-of-service conditions via crafted DNS queries or responses, memory/CPU exhaustion, or cache poisoning through DNSSEC validation bypasses. The vulnerabilities affect multiple versions of BIND 9 across stable and development branches, with no workarounds available. While ISC reports no known active exploitation, public reproduction tests exist, increasing the risk of future exploitation.

1mo ago · hacker-news

The China-aligned state-sponsored threat actor FamousSparrow has been deploying a new modular C++ backdoor named SparroWocky in targeted cyber espionage attacks across Latin America since at least August 2025. The malware, which replaces the group's previous SparrowDoor implant, supports command execution, file exfiltration, periodic screenshots, and acts as a TCP proxy. It uses anti-analysis techniques and integrates open-source tools like Mbed TLS, MinHook, and COFF Loader for secure communications, evasion, and in-memory plugin execution. Targets include governmental entities in Argentina, Ecuador, Guatemala, Honduras, Panama, Peru, Puerto Rico, and Venezuela, with 90% of observed activity focused on the region.

1mo ago · hacker-news

A critical heap overflow vulnerability (CVE-2026-81642) exists in Unbound DNS resolver versions up to and including 1.26.0, which can be exploited remotely by an attacker controlling a malicious DNS zone to trigger denial of service or potentially achieve remote code execution. The vulnerability resides in the DNSSEC validator when processing a DNSKEY record with a compression pointer pointing into the record's own data. A second high-severity vulnerability (CVE-2026-82717), a heap corruption in CNAME synthesis, also affects the same versions and could lead to remote code execution under specific conditions. Both flaws are patched in Unbound 1.26.1.

1mo ago · bleeping-computer

The FBI has seized the domains of NightmareStresser, a long-running DDoS-for-hire ('booter') service that enabled users to launch large-scale distributed denial-of-service attacks using compromised IoT devices and routers. The service, which claimed to be the '#1 online IP booter,' was used in hundreds of thousands of attacks since 2022, targeting networks globally with attack volumes up to 200 Gbps across multiple layers. The takedown was part of Operation PowerOFF, an international law enforcement effort targeting criminal DDoS infrastructure, which has previously dismantled similar services and led to multiple arrests and domain seizures worldwide.

1mo ago · bleeping-computer

Cisco has warned of active exploitation of a maximum-severity zero-day vulnerability, CVE-2026-76460, in its Identity Services Engine (ISE) and ISE-PIC software. The flaw allows remote attackers to bypass authentication by exploiting insufficient controls on an API endpoint, enabling unauthorized access to the web-based management interface. No workarounds exist, and Cisco strongly recommends applying security updates immediately. The Cybersecurity and Infrastructure Security Agency (CISA) has added the vulnerability to its Known Exploited Vulnerabilities catalog and mandated federal agencies to patch within three days.

1mo ago · bleeping-computer

The China-linked threat actor FamousSparrow has been conducting espionage campaigns against government organizations in Latin America since mid-2025, using a new modular C++ backdoor named SparroWocky. This malware replaces the group's previously used SparrowDoor backdoor and is deployed via DLL side-loading, with payloads RC4-encoded in .dat files. SparroWocky features advanced evasion techniques, including thread creation hooking to spoof legitimate Windows functions, in-memory code manipulation, and anti-analysis tricks, enabling command execution, file manipulation, screenshot capture, and proxy capabilities. The malware establishes persistence via a Windows service or registry key and communicates with C2 servers over ports 443 or 8080, sometimes through HTTP and SOCKS5 proxies.

1mo ago · unit42

This report details an analysis of the Atomic macOS (AMOS) stealer malware, observed in a lab environment on August 5, 2026. AMOS stealer is distributed via malicious websites and cracked software campaigns, using social engineering to trick users into pasting malicious commands into Terminal. The infection chain involves downloading a Zsh script that retrieves and executes a Mach-O binary, which establishes persistence and collects sensitive data including credentials, browser data, cryptocurrency wallets, and messaging app data. Exfiltration occurs via HTTP POST requests to command and control servers, with infrastructure and indicators frequently changing, indicating active development and evasion tactics.

1mo ago · hacker-news

Three distinct threat groups—NightEagle, Hacking Cat, and Toy Ghouls—are targeting Russian enterprises using backdoors, ransomware, and wipers. NightEagle exploits vulnerabilities like CVE-2019-0708 and CVE-2020-0688 to deploy the GhostContainer backdoor on Microsoft Exchange servers, enabling code execution and lateral movement. Hacking Cat, a pro-Ukrainian hacktivist group, uses Gorilla RAT and multiple variants of Monkey ransomware written in different languages, along with wiper malware Nemo Wiper, often in collaboration with other groups. Toy Ghouls has shifted to custom tools, deploying a new backdoor called Bird Agent that uses HiveMQ MQTT or Element (Matrix) for C2 communication, delivered via WinRM using tools like Evil-WinRM.

1mo ago · hacker-news

Security researchers at Forever Security demonstrated a method to hijack AI assistants in multiple Chromium-based browsers and extensions by exploiting browser extension permissions to intercept trusted communication channels. The technique leverages two common extension permissions—content modification and network request manipulation—to inject malicious code into trusted AI service pages, enabling unauthorized access to sensitive capabilities such as file reading, screenshot capture, and AI agent control. While no active exploitation in the wild has been observed, the research highlights critical design flaws in how AI agents are integrated into browsers, with CVE-2026-0628 (Chrome) and CVE-2026-55945 (Edge) officially recognized and patched, while similar issues in Comet, Opera Neon, and Claude in Chrome remain less formally addressed.