Koffei

Daily CyberNews · 04/11/2026

‹ 03/11/202605/11/2026 ›
Today in 30 seconds
  • 🔴 Patch WordPress Core: CVE-2026-119977 is exploited, in KEV, deadline Nov 6.
  • 🔴 Update cPGuard to 5.88.25: its update channel pushed a backdoor to Linux servers.
  • 🔴 Update Node-RED on SIMATIC IoT2050: CVE-2026-58115 is exploited, CVSS 10.0.
  • 🔴 BTCPay Server v2.4.1 hot wallets are being drained: move funds to a fresh install.
  • 🔴 Remove the CryptoPayment Gateway plugin and rotate wallet keys: CVE-2026-81648 has no fix.

Vulnerabilities 11

🔴 Exploited or active

Siemens SIMATIC IoT2050 / Node-RED — actively exploited, CVSS 10.0

CVE-2026-58115Siemens SIMATIC IoT2050CVSS 10.0

WhatA missing-authentication flaw in the Node-RED HTTP interface on SIMATIC IoT2050 Advanced allows unauthenticated code execution; exploitation attempts seen since Oct 30.

DoUpdate to Node-RED 4.3.4.1 or later per SSA-834709, or disable or harden the Node-RED HTTP interface meanwhile.

Details and sources
CVE-2026-58115 is a missing-authentication flaw in the Node-RED HTTP interface on SIMATIC IoT2050 Advanced devices, letting an unauthenticated remote attacker run arbitrary code. VulnCheck has observed in-the-wild exploitation attempts against its honeypots since October 30; Siemens advisory SSA-834709 fixes it in Node-RED 4.3.4.1+ (interim mitigation: disable/harden the Node-RED HTTP interface). The Hacker News (source)
🔴 Exploited or active

WordPress "CryptoPayment Gateway" plugin — actively exploited, no patch

CVE-2026-81648CryptoPayment Gateway pluginCVSS 10.0no fix

WhatAn unauthenticated AJAX flaw in this WordPress plugin exposes crypto wallet credentials and is under active exploitation attempts.

DoRemove the plugin, rotate wallet keys and audit the site for compromise.

Details and sources
CVE-2026-81648 (CVSS 10.0) is an unauthenticated AJAX flaw exposing crypto wallet credentials; Defused Cyber/WPScan report active in-the-wild exploitation attempts. The plugin was pulled from the WordPress directory on Aug 28, 2026 and has no fix — site owners are told to remove it, rotate wallet keys, and audit for compromise. The Hacker News (source)
🟠 High impact

Barracuda — two flaws chainable to root

CVE-2026-116606CVE-2026-111560BarracudaCVSS 9.3

WhatAn authorization bypass and a command injection in Barracuda appliances can be chained to get root through the admin interface; no exploitation confirmed.

DoConfirm Barracuda appliances run the latest firmware and patches.

Details and sources
CVE-2026-116606 (CVSS 9.3, pre-auth authorization bypass, Email Security Gateway) and CVE-2026-111560 (CVSS 8.6, command injection, affecting Email Security Gateway, Load Balancer ADC, Message Archiver, WAF Control Center, Web Application Firewall and Web Security Gateway) can be chained to obtain root access via the admin interface. No in-the-wild exploitation confirmed yet; Barracuda advises customers to confirm they are on the latest firmware/patches. The Hacker News (source)
🟠 High impact

Linux kernel IPv4/TCP use-after-free, CVSS 9.8

CVE-2026-98432Linux kernel TCPCVSS 9.8

WhatA race condition in the IPv4 TCP path since Linux 4.4 allows remote root; Anthropic says an internal Claude model wrote an exploit, but no exploitation is confirmed.

DoInstall Red Hat's Nov 1 kernels for RHEL 8.10/9.8/10.2 and reboot; Ubuntu fixes are still in -proposed.

Follow-ups: 17/11
Details and sources
CVE-2026-98432 is a race condition in the kernel's IPv4 TCP path (present since Linux 4.4, MPTCP-reachable); Red Hat initially rated it 5.9 before raising it to 9.8/Critical. Red Hat shipped fixed kernels Nov 1 for RHEL 8.10/9.8/10.2 (reboot required); Ubuntu fixes for 22.04/24.04/26.04 are in the -proposed pocket. No confirmed in-the-wild exploitation and not yet in CISA's KEV catalog. Notably, Anthropic disclosed on Oct 27 that an unreleased, internal Claude research model wrote a working remote-root exploit for the flaw (not published) — Anthropic's own claim, not independently verified. The Hacker News (source)
🔴 Exploited or active

WordPress Core — CISA confirms active exploitation, CVSS 8.1

CVE-2026-119977WordPress CoreCVSS 8.1in KEV

WhatA race condition in WordPress Core upgrade staging paths gives unauthenticated RCE; Wordfence logged over 18,500 exploitation attempts in 21 hours.

DoPatch WordPress Core; the federal KEV deadline is Nov 6.

Details and sources
CVE-2026-119977 is a time-of-check/time-of-use race condition in predictable, web-accessible staging paths used during full-package Core upgrades, letting an attacker reach remote code execution by hitting a brief window mid-update — no account needed. CISA added it to the KEV catalog, giving FCEB agencies until Nov 6, 2026 to patch; Wordfence logged over 18,500 exploitation attempts against staged installer paths in the first 21 hours after disclosure. CISA · The Hacker News (source)
🟠 High impact

Google patches critical Android heap overflow

Added after publishingCVE-2026-58868Android 14 to 17Critical

WhatA heap buffer overflow in FrameDecoder.cpp affects Android 14 through 17; no active exploitation reported.

DoInstall the November 2026 Android update with patch level 2026-11-01 or later.

Details and sources
CVE-2026-58868 is a heap buffer overflow in FrameDecoder.cpp's onOutputReceived function, rated Critical by Google, affecting Android 14 through 17. No active exploitation reported and not in CISA's KEV catalog as of Nov 3. Fixed in the November 2026 Android Security Bulletin (security patch level 2026-11-01 or later); rollout timing depends on device makers and carriers. Google · The Hacker News (source)
🟠 High impact

Ruby on Rails — critical Active Storage flaw, CVSS 9.8, no CVE yet

Added after publishingRuby on Rails Active StorageGHSA-f3w6-w445-fwh9CVSS 9.8

WhatAn unauthenticated crafted upload to apps using ImageMagick lets an attacker write arbitrary files, and Disk service apps can leak secrets; one real attack chain was reported.

DoUpgrade Active Storage to 8.1.4.1 or 8.0.5.2; pre-8.0 series get no fix.

Details and sources
Tracked as GHSA-f3w6-w445-fwh9, this flaw lets an unauthenticated attacker write arbitrary files on a server by uploading a crafted file to an app that processes uploads with ImageMagick; apps using the Disk service can also have any readable file (such as secret_key_base or third-party credentials) read back, potentially enabling RCE or lateral movement. Rails says one real attack chain was reported to it. Affects Active Storage before 8.0.5.2 and 8.1 before 8.1.4.1 (pre-8.0 series are unsupported and won't get a fix); patched in 8.1.4.1 and 8.0.5.2. Rails · The Hacker News (source)
🟠 High impact

Xen hypervisor — guest-to-host VM escape, XSA-523

Added after publishingCVE-2026-102703CVE-2026-102704Xen XSA-523

WhatA flaw in Xen's grant copy operation lets a malicious guest VM read and write other VMs' memory, including dom0; no attacks observed.

DoPatch Xen 4.17 to 4.19, XCP-ng 8.3 to xen-4.17.7-4.2.xcpng8.3 or Qubes dom0 to Xen 4.19.6-3, then reboot.

Details and sources
CVE-2026-102703 and CVE-2026-102704 are flaws in Xen's "grant copy" operation, which reuses permission checks across a batched memory-page copy without always rechecking — letting a malicious guest VM read and write the memory of other VMs on the same host, including dom0. Vates/XCP-ng rates the impact Critical ("a single compromised guest can break out of its isolation and take over the whole host"); no CVSS given, no known mitigation besides patching, no observed in-the-wild attacks. Affects all Xen 4.6+ (4.5 and earlier unaffected). Fixed upstream for Xen 4.17–4.19.x; no patch for 4.6–4.16 (4.16 support ended Dec 2024); XCP-ng 8.3 fixed in xen-4.17.7-4.2.xcpng8.3 (reboot required); Qubes OS 4.3 fixed in dom0 Xen 4.19.6-3 (also requires a dom0 restart, and Anti Evil Maid users must reseal their passphrase). Xen Project · The Hacker News (source)
🟠 High impact

Chrome 157 — 213 security fixes, 7 rated Critical

Added after publishingChrome 157CVE-2026-1212297 critical

WhatChrome 157 fixes 213 security issues, including an ANGLE memory-corruption bug allowing code execution outside the sandbox; no zero-day exploitation confirmed.

DoUpdate Chrome to version 157 on desktop and Android.

Details and sources
Released to Stable (Windows/Mac/Linux and Android) on Nov 3. Among the critical flaws: CVE-2026-121229, a memory-corruption bug in ANGLE that could let a remote attacker run code outside the sandbox via a crafted HTML page (affects Chrome for Android); plus a use-after-free in Prerender (CVE-2026-121035) and a buffer overflow in the Dawn/WebGPU component (CVE-2026-121156). No confirmation of active/zero-day exploitation among the critical flaws. Malwarebytes (source)
🟠 High impact

APT (Debian/Ubuntu package manager) — MITM-to-root flaw

Added after publishingCVE-2026-109518APTDebian and Ubuntu

WhatAn attacker who can modify HTTP package downloads, such as a malicious mirror or proxy, can swap a verified package and run code as root.

DoInstall the patched APT builds from Debian and Ubuntu; there is no workaround.

Details and sources
CVE-2026-109518 (Ubuntu CVSS 4.0 score: 7.7/High; APT's own developers call it Critical) lets an attacker who can modify HTTP package-download traffic — e.g. a malicious mirror or proxy — swap a verified package for their own using HTTP redirects/retries, with the resulting code running as root. Affects APT's default retry behavior from Debian 12 / Ubuntu 22.04 onward (older releases exposed only if Acquire::Retries was manually enabled); no in-the-wild exploitation reported; reported by researcher Max Justicz. Fixed: Debian has released patched APT builds (including a new 2.2.5 for EOL Debian 11); Ubuntu shipped fixes Nov 4 for 22.04, 24.04, 26.04 and 26.10, with more (incl. Ubuntu Pro) to follow. No workaround for systems that can't update yet. The Hacker News (source)
🟠 High impact

SonicWall GMS — two flaws, no patch coming (EOL)

Added after publishingCVE-2026-109593CVE-2026-109594SonicWall GMSEOL

WhatSonicWall GMS 9.5.2 and earlier has an unauthenticated file read and a SQL injection that will never be patched; separate attacks target CVE-2026-102255.

DoMigrate from GMS to SonicWall Network Security Manager, and make sure the patch for CVE-2026-102255 is applied.

Details and sources
CVE-2026-109593 (CVSS 9.1, unauthenticated file read exposing internal GMS files) and CVE-2026-109594 (CVSS 6.8, post-auth SQL injection) affect SonicWall Global Management System 9.5.2 and earlier. SonicWall says the product "has reached End-of-life" and "no remediation or security update will be released" — affected organizations are told to upgrade to SonicWall Network Security Manager (NSM) instead. Separately (not these two bugs), Previdian telemetry shows ongoing attack attempts — 2,541 since Oct 9, up to 30 on Nov 3 alone — against a different, already-patched SonicWall flaw, CVE-2026-102255 (CVSS 10.0, pre-auth SSRF). The Hacker News (source)

Breaches / Incidents 6

🔴 Exploited or active

BTCPay Server — hot-wallet theft

BTCPay Serverhot walletsv2.4.1 and earlier

WhatAttackers are draining Bitcoin and Litecoin from BTCPay Server hot wallets on v2.4.1 and earlier using a new method.

DoTreat affected servers and secrets as compromised and move funds to a fresh install on the current 2.4.7 release.

Details and sources
Attackers are draining Bitcoin and Litecoin from BTCPay Server hot wallets on v2.4.1 and earlier, using a different method than the credential-exposure flaw fixed in v2.4.2 (Aug 7, 2026); current release 2.4.7 has no credible reports of drains. Affected operators are advised to treat the whole server and its secrets as compromised and move funds to a fresh install. The Hacker News (source)
🔴 Exploited or active

cPGuard (OPSSHIELD) update-channel compromise

cPGuardOPSSHIELDupdate channel hijacked

WhatAn attacker with stolen credentials pushed a malicious script through cPGuard updates to Linux servers on 5.88.12 to 5.88.19, installing a backdoor and stealing credential files.

DoUpdate cPGuard to 5.88.25 or later, which removes the backdoor and leftover files.

Details and sources
An attacker used stolen credentials to access cPGuard's own update system (02:06 UTC, Nov 2) and pushed a malicious script to customer Linux servers, versions 5.88.12–5.88.19, installing a backdoor and harvesting /shadow, /passwd and other credential files. Fixed in 5.88.25+, which also removes the backdoor and leftover files. The Hacker News (source)
🟠 High impact

Chainstack breach — customer database and node credentials exposed

Chainstacknode credentialspossible ARTEX link

WhatAn attacker moved from Chainstack's development environment into production for five days, exposing a customer database and systems holding node credentials.

Details and sources
The blockchain infrastructure provider says an attacker reached a development environment on Oct 26, 2026, escalated into parts of production, and was cut off on Oct 31. Exposed: a customer database (org names, member emails, password hashes) and systems holding node credentials — Chainstack found no evidence the credentials were read but is treating them as compromised. The company is assessing a possible, unconfirmed link to a large-scale AI-powered campaign resembling the China-linked ARTEX cluster. The Hacker News (source)
🟡 Worth knowing

Philippine Department of Migrant Workers — ~68GB of OFW records copied

Philippines DMW68GBmigrant worker records

What27 SQL Server backups with old migrant-worker records were copied from an exposed Singapore-hosted server in September; personal data content is unconfirmed.

Details and sources
Researchers found that 27 Microsoft SQL Server database backups carrying old POEA/DMW migrant-worker records were copied from an exposed, Singapore-hosted server in under 24 minutes on Sept 6, 2026; the find was disclosed Oct 28. The Philippine National Privacy Commission is investigating; no attacker attribution yet, and it's unconfirmed whether the backups held personal data or whether anyone else accessed them. GulfNews (source) — also covered by Dark Reading (source)
🟠 High impact

mSecure password manager — vendor warns of "suspicious activity"

Added after publishingmSecurepassword managersuspicious activity

WhatmSecure's maker warned current and former customers of unspecified suspicious activity linked to the app without saying what data was exposed.

DoChange passwords stored in mSecure, reset or re-enroll 2FA and remove passkeys.

Details and sources
mSeven Software, maker of the mSecure password manager, emailed current and former customers on Nov 3 flagging unspecified "suspicious activity linked to the app." It says it has engaged outside cybersecurity experts and "took steps to mitigate the issue," and tells both current and former customers to change passwords, reset or re-enroll 2FA, and remove passkeys — mSecure's own documentation warns someone may already have backed up vault contents, so a password change alone may not be enough. mSeven still hasn't said when the activity occurred, which systems were involved, or whether any customer data (including stored vault secrets) was actually exposed. The Hacker News (source) · Malwarebytes (source)
🟠 High impact

French Blood Establishment (EFS) refrigeration tampering

Added after publishingEFSFrancerefrigeration tamperingCoolworld

WhatAttackers changed temperature setpoints on refrigeration used by France's blood service via its cooling provider; a technician caught it in time.

Details and sources
Overnight Nov 1–2, 2026, attackers reportedly changed temperature setpoints (from around −30°C to 110°C) on refrigeration equipment used by EFS, France's blood service, after targeting its cooling-service provider, Coolworld; facilities in Bourgogne-Franche-Comté were among those affected. Donor samples and some transfusion-bound frozen plasma were involved, but an on-call technician caught the abnormal readings in time — France's Health Ministry says there were no confirmed operational consequences, no confirmed destroyed plasma, and no attribution yet. The outlet is careful to note this is a separate case from October's French supermarket refrigeration attacks (which hit Danfoss/Copeland/Dixell controllers) — similar behavior, but no public evidence linking the two. Cybersecurity News (source)

Malware / Ransomware 6

🔴 Exploited or active

C3RB3R ransomware targets Synology NAS

C3RB3RSynology DSMCVE-2024-10441

WhatC3RB3R ransomware encrypts unpatched Synology NAS devices, possibly via CVE-2026-13684 and CVE-2026-13639 as well as older DSM flaws.

DoPatch Synology DSM for advisories SA-24:20 and SA-26:13.

Details and sources
Encrypts files with a .L0CK3D extension and drops a "read-me3.txt" note, exploiting unpatched DSM installs. Synology's Nov 2024 advisory covers CVE-2024-10441 (CVSS 9.8, RCE in the system plugin daemon); Qihoo 360 separately ties recent activity to September 2026 flaws CVE-2026-13684 and CVE-2026-13639 (CVSS 9.8 each, under Synology advisory SA-26:13). The Hacker News (source)
🟠 High impact

VOLTA-AUTODOXER infostealer pipes stolen data through Claude

VOLTA-AUTODOXERinfostealerClaude

WhatA Windows infostealer sends stolen credentials, cards and seed phrases to Claude, Kimi or Gemini to build a ranked victim dossier, exfiltrated via Telegram.

Details and sources
A Windows infostealer (tracked by Cisco Talos) that harvests browser credentials, cookies, payment cards, crypto seed phrases and more, then sends the haul through the OpenRouter gateway to Claude Opus 4.6 (falling back to Moonshot's Kimi K2, then Google's Gemini 2.5 Flash) to auto-generate a ranked "financial value" and OPSEC dossier on the victim. Data and the AI-generated report are exfiltrated via Telegram. The Hacker News (source)
🟠 High impact

Aur0ra ransomware (tracked by Sophos as GOLD OGRE) directs AI coding agents through intrusions

Aur0raGOLD OGREAI coding agents

WhatAur0ra ransomware operators used Cursor AI agents to map Active Directory, hunt credentials and move laterally, reaching domain admin before deploying ransomware.

Details and sources
Sophos recovered Cursor AI agent artifacts showing operators (Russian-language prompts) handing the agent VPN/proxy access and objectives; the agent then mapped Active Directory, hunted for stored credentials, and moved laterally "with little input," reaching domain-admin access on some networks before humans took over for ransomware deployment. Attacks ran through Cursor's Composer model plus models from Anthropic, xAI and Moonshot. Sophos via Help Net Security (source)
🟠 High impact

3AM/ThreeAM ransomware hits UK school estate

ThreeAM3AM ransomwareUK schools

WhatA 3AM affiliate breached 47 systems across UK school AD domains, abused cross-domain trusts and deleted backups before encrypting about 37.

Details and sources
An affiliate breached 47 systems across separate AD domains (about 37 encrypted), using Mimikatz, DPAPI credential decryption, and dnscmd/netdom to abuse cross-domain trusts, then wbadmin to delete backups before encrypting. Bitdefender assesses the intrusion as a possible "hybrid" operation where command timing suggests an AI agentic loop may have suggested some commands — not confirmed as autonomous. Bitdefender via Hackread (source)
🟡 Worth knowing

SysGoRAT hits government Linux servers

SysGoRATCL-UNK-1211government Linux

WhatA Go-based RAT with WebSocket C2 hit government Linux systems, with attacker scripts showing signs of AI-generated code.

Details and sources
A custom Go-based, cross-platform RAT (tracked as CL-UNK-1211) with in-memory reverse shells and WebSocket C2, used against Linux systems at government and state-affiliated organizations. Palo Alto Unit 42 says several of the attackers' PHP/shell scripts contain conversational reasoning notes (and at least one syntax error) consistent with AI-generated or AI-agent-deployed code. Palo Alto Networks Unit 42 via The Hacker News (source)
🟡 Worth knowing

Haiou "prompt adware" hijacks OpenAI Codex via natural language, not code

Added after publishingHaiouprompt adwareOpenAI Codex

WhatA jailbreak bundle persistently edits Codex config files so greetings make the agent promote an opaque LLM gateway; it spreads via GitHub and Codex-X.

Details and sources
Check Point Research coined the term for this jailbreak bundle, which uses prompt-injection techniques (role-play, reframing requests as CTF scenarios) to get past Codex's safety guardrails, then persistently edits its config/global instruction files so the change survives future sessions. Common greetings ("hi," "hey," "Seagull") then trigger the agent to promote an opaque, OpenAI-compatible LLM gateway at zz.aiapi2025[.]top; the bundle also installs pentesting/password-cracking/reverse-engineering "skills." Distributed via a GitHub repo ("haiou-codex," 403 stars/77 forks, published July 10, 2026) and bundled into Codex-X, a Chinese GUI config tool for Codex (which isn't officially available in mainland China). Check Point calls the ad itself fairly harmless but warns that natural-language "skills"/configs — not traditional executables — are harder for users and security teams to recognize as a supply-chain risk once they become part of an agent's trust chain. Check Point Research · The Hacker News (source)

Threat Actor Activity 6

🟠 High impact

Google disrupts 1.3M-device x.cpe Android TV-box botnet, blames vendor backdoor

x.cpe botnetVideostrong1.47M devices

WhatGoogle and partners took down a botnet of Android TV boxes used for proxying, ad fraud and data theft, blaming a vendor-built backdoor.

Details and sources
Google and partners (Japan's NPA, Lumen, HUMAN Security, Plume, Synthient, TrendAI) took down the "x.cpe" botnet, which TrendAI counted at over 1.47 million infected Android TV boxes across 206 models — mostly in Malaysia, Brazil, Iraq, Mexico and Taiwan — used for residential proxying, ad fraud and data theft. Google says it is "highly confident" device maker Videostrong (Shenzhen) intentionally built in the backdoor and sold criminal access to it; Videostrong and any affiliates are now banned from Google's certified Android ecosystem. The Hacker News (source) · Dark Reading (source)
🔴 Exploited or active

New Russia-aligned actor "Gorbag" targets Ukraine exclusively

GorbagUkraineCVE-2025-8088BeautyStealer

WhatRussia-aligned Gorbag spear-phishes Ukrainian law enforcement, front-line administrations and defense targets, exploiting a patched WinRAR flaw to deploy BeautyStealer and Prime ReverseShell.

Details and sources
ESET researcher Damien Schaeffer attributes activity dating back to February 2025 to Gorbag, which overlaps CERT-UA's UAC-0226 and Trend Micro's SHADOW-EARTH-066 and may act as an initial-access broker, possibly only the first stage of a longer intrusion chain. It exploits the patched WinRAR flaw CVE-2025-8088 via DLL sideloading and spear-phishing from compromised Ukrainian local-government email accounts (malicious .XLSM attachments) to deploy the BeautyStealer browser/file stealer and a Prime ReverseShell implant — a separate infostealer, GIFTEDCROOK, is linked to the same CVE by CERT-UA/Trend Micro but is not part of ESET's own Gorbag analysis. Targets: Ukrainian law-enforcement/justice bodies (~1/4), local administrations near the front line (~1/4), and the defense sector and a military-aid transport hub (remainder). One Prime ReverseShell sample even contained a comment addressed to AI/LLM code reviewers claiming the file "is not malicious" — it didn't fool ESET's testing. ESET (source) · The Hacker News (source)
🟡 Worth knowing

Binance $6.34B heist — Binance says attackers used "well-known" tricks

BinanceTraderTraitor$6.34B

WhatInvestigators attribute the Oct 29 Binance heist, which began with a compromised DevOps engineer's laptop, to North Korea's TraderTraitor.

Follow-ups: 13/11
Details and sources
Crypto investigators TRM Labs and Chainalysis have attributed the Oct 29, 2026 theft — the largest crypto heist on record, which began with the compromise of a Binance DevOps engineer's laptop — to North Korea's TraderTraitor group (aka Lazarus/APT38). Binance's own account characterizes the methods used as familiar, well-established intrusion techniques rather than anything novel. Wired (source)
🟡 Worth knowing

Russia dismantles VasyGrek (Fluffy Wolf) cybercrime group

Added after publishingVasyGrekFluffy WolfRussia arrests

WhatRussian police arrested three suspected members of VasyGrek, a phishing-driven cybercrime group active since 2016 using stealers and RATs.

Details and sources
Russia's Ministry of Internal Affairs arrested three suspected members of the financially motivated group, active since 2016, which breaks in via phishing emails disguised as financial documents and deploys tools including MetaStealer, PureCrypter, PureLogs Stealer, WarZone RAT, XMRig and PureHVNC. Authorities searched suspects' homes/offices and tied the group to a Sept 2025 theft of 660,000 rubles (~$8,000) from one victim organization; Kaspersky logged over 100 attempted VasyGrek attacks against targets in Russia, Belarus and Kazakhstan between January and September 2026. MVD · The Hacker News (source)
🔴 Exploited or active

SlippyNunchucks (China-aligned) hijacks Sogou Input Method's update channel

Added after publishingSlippyNunchucksSogouCVE-2026-120422DarkVigil

WhatA China-aligned actor tampered with DNS for Sogou's update component to deliver the DarkVigil backdoor to Chinese-speaking targets across Asia since July 2026.

DoUpdate Sogou Input Method to v16.10, which moves to HTTPS.

Details and sources
ESET (researcher Anh Ho) tracks this distinct intrusion set — previously miscast as overlapping with Mustang Panda — active since at least Sept 2023. It tampered with DNS resolution for Sogou's SGTool.exe vocabulary-promotion component (normally fetched over plain HTTP from config.pinyin.sogou[.]com), redirecting it to attacker servers that returned a malicious config, which triggered DLL sideloading via a legitimate Adobe executable to launch the modular DarkVigil (FDMTP) backdoor. Tencent fixed the underlying flaw, CVE-2026-120422, in Sogou v16.10 by moving to HTTPS. The actor pivoted to Sogou three days after a prior compromise of the QuickFox VPN tool was cleaned up; targets include Chinese-speaking individuals and organizations in Hong Kong, Thailand, Japan, Vietnam, Oman and Sri Lanka since July 2026, including surveillance of Chinese-language online betting operators — ESET assesses this as nation-state surveillance likely serving Chinese government interests. ESET · The Hacker News (source)
🔴 Exploited or active

APT29-linked "Poisoned Source" watering-hole campaign hits iOS/Android

Added after publishingAPT29Poisoned Sourcewatering holeiOS and Android

WhatAn APT29-linked watering-hole campaign serves N-day exploit chains to iOS and Android visitors of compromised sites, stealing cookies and full device data.

Details and sources
French firm Sekoia (researchers Amaury-Jacques Garçon and Félix Aime) says APT29 is "said to be" behind a watering-hole campaign discovered mid-October 2026, using three vectors: malicious JavaScript injected into compromised legitimate sites, lookalike domains (e.g. "america250[.]holiday") targeting visitors in the US, Ukraine, Czechia and Bulgaria, and a supply-chain compromise of Ukrainian analytics service Plerdy SEO Tracker (malicious code live Aug 27–Sept 3, 2026). Injected JS fingerprints the visitor's device and serves a platform-specific N-day exploit chain: iOS 16.7 victims get cookie-theft JS via CVE-2025-43529/CVE-2024-27834; iOS 18.5 victims get the "DarkSword" chain for full data exfiltration (messages, Wi-Fi passwords, Keychain); Android/Chrome 139–145 (excluding 143) victims get the AndroSnitch stealer via CVE-2026-3910/CVE-2025-27038 — notably, the Android chain deliberately skips Samsung Galaxy S25 Ultra devices. Compromised/targeted sites include a US chemical-industry trade group, a far-right European political party, and a Bulgarian military-linked site; Sekoia suspects the campaign targets specific high-value individuals given the Russia-Ukraine war context, and draws a parallel to a 2024 APT29-suspected campaign using a similarly named payload (ANDROSNATCH). Sekoia · The Hacker News (source)
‹ 03/11/202605/11/2026 ›