Koffei

Daily CyberNews · 04/09/2026

‹ 03/09/202607/09/2026 ›
Today in 30 seconds
  • 🔴 Patch the seven new KEV flaws and check for unauthorized SSH key changes.
  • 🔴 Shai-Hulud npm worm now hunts 469 credential locations and self-propagates.
  • 🟠 CrowdStrike Falcon users: disable the Office macro-removal policy until FalconFlank is fixed.
  • 🟠 Restrict TCP 43210/43211 on Nexus 9000: CVE-2026-20212 gives unauthenticated root RCE.
  • 🟠 Check Virtualizor hosts for rogue root SSH keys and a hidden proxyuser.

Vulnerabilities 3

🟠 High impact

Cisco Nexus 9000 critical RCE

CVE-2026-20212Cisco Nexus 9000CVSS 9.8not exploited

WhatUnauthenticated attackers can run code as root on 10 Silicon One Nexus 9000 models via unrestricted TCP ports 43210/43211 in the default VRF.

DoRestrict access to TCP ports 43210 and 43211 in the default VRF on affected Nexus 9000 switches.

Details and sources
CVE-2026-20212 (CVSS 9.8) lets an unauthenticated remote attacker run code as root on 10 Silicon One-based Nexus 9000 switch models by hitting unrestricted TCP ports 43210/43211 in the default VRF; Cisco says it's not aware of exploitation yet, but the flaw can also crash devices via S1HAL. The Hacker News
🔴 Exploited or active

CISA KEV +7, actively exploited

CISA KEV7 flawsSonicWallKestraLiteLLM

WhatExploited SonicWall SMA 1000, Switchvox, Artifactory, Starlette, Kestra and LiteLLM flaws are used for reverse shells, crypto mining, API key theft and SSH-key persistence.

DoPatch these products and check hosts for unauthorized SSH key changes.

Details and sources
SonicWall SMA 1000 SSRF + post-auth OS command injection, Sangoma Switchvox unauthenticated SQLi/RCE, JFrog Artifactory auth bypass, Starlette request smuggling, Kestra OSS unauthenticated command injection, and Berri LiteLLM MCP auth bypass. Attackers are using them to drop reverse shells, mine crypto, steal API keys, and persist via SSH key changes. CISA · The Hacker News
🟠 High impact

CrowdStrike Falcon local-priv-esc PoC ("FalconFlank")

FalconFlankCrowdStrike Falconzero-dayno fix

WhatAn unpatched zero-day abuses Falcon's Office macro remediation for privilege escalation on patched Windows 11 25H2 and Server 2025; similar PoCs target Kaspersky and Defender.

DoDisable the affected Falcon Office macro-removal policy, as CrowdStrike recommends.

Details and sources
unpatched zero-day abusing Falcon's Office macro-remediation feature for privilege escalation on fully patched Windows 11 25H2 / Server 2025; CrowdStrike has no fix yet and recommends disabling the affected Office macro-removal policy. Researcher released similar PoCs against Kaspersky and Microsoft Defender the same day. The Hacker News

Breaches / Incidents 2

🟠 High impact

Thomson Reuters C-Track court software breach

Thomson ReutersC-Trackcourt records

WhatAttackers accessed backup court files with SSNs, licenses and sealed records from March to June 2026, affecting courts in 11 US states and elsewhere.

Details and sources
unauthorized access to backup court files (SSNs, driver's license numbers, DOBs, medical/health insurance data, some sealed/redacted records) between March 1–June 29, 2026, discovered June 30 but only disclosed September 2. Affects courts in 11 U.S. states plus the U.S. Virgin Islands, Ontario, and Minnesota. No confirmed fraud yet; total individuals affected and intrusion method still undisclosed. The Hacker News
🟠 High impact

IDScan.net / "Nexus" dark-web breach, 153M+ driver's licenses

IDScan.netNexus153M driver's licensesFBI

WhatA new dark-web marketplace sells an alleged 153M driver's license scans plus millions of other IDs sourced from IDScan.net; the FBI is investigating.

Details and sources
a new dark-web marketplace is selling an alleged 153M scanned driver's licenses plus 10M ID cards, 3M travel documents and 579K medical cards (front/back, IR/UV scans, timestamps) sourced from identity-verification provider IDScan.net. FBI's New Orleans field office is investigating; IDScan.net says it's looking into it. Malwarebytes

Malware / Ransomware 5

🔴 Exploited or active

Shai-Hulud worm expands credential-hunting to 469 locations

Shai-Huludnpm worm469 locations

WhatThe npm supply-chain worm now sweeps 469 credential locations, including CI/CD configs, cloud creds and AI-tool configs, and self-propagates with stolen publishing tokens.

Details and sources
up from 189, the npm/dev-tooling supply-chain worm now sweeps shell histories, CI/CD configs, cloud creds, AI-tool configs and package-manager auth files; stolen publishing tokens let it self-propagate through trusted package channels. The Hacker News
🟠 High impact

BraZetsu malware / "Infected Marketplace"

BraZetsuExilwareBanco de Infects

WhatAn AI-assisted Python Windows framework profiles infected hosts over WebSocket C2 and resells access for about $5.80, fueling attacks in Iberia and LatAm.

Details and sources
AI-assisted, Python-based Windows framework from threat cluster Exilware that profiles infected hosts (browser data, certs, screenshots, Brazilian CNAB financial files) over WebSocket C2 and resells access for ~$5.80 via a marketplace ("Banco de Infects"), fueling follow-on attacks across Iberia/LatAm finance, e-commerce and critical infrastructure. The Hacker News
🟠 High impact

Fake installer campaign disabling Windows Update / weakening Defender

Silver Foxfake installersChina

WhatSpoofed vendor sites push installers that disable Windows Update, add Defender exclusions, delete shadow copies and use non-standard C2 ports.

DoHunt for new Defender exclusions, disabled Windows Update and deleted shadow copies.

Details and sources
attributed with moderate confidence to Chinese cluster Silver Fox (Yinhu); spoofed vendor download sites (.com.cn/.hl.cn) push wrapper installers that disable Windows Update, add Defender exclusions, delete shadow copies, and set up non-standard C2 ports. Victims span healthcare, manufacturing, gaming, government and education, mostly in China. The Hacker News
🟠 High impact

BGP hijack delivers trojanized Virtualizor/Softaculous update

VirtualizorSoftaculousBGP hijackAlbaHost

WhatRogue BGP routes on August 28-30 delivered a tampered Virtualizor update adding root SSH keys, a hidden proxyuser and systemd persistence; AlbaHost lost 5 hypervisors.

DoCheck Virtualizor hosts for unknown root SSH keys, a proxyuser account, Java 17 payloads and new systemd services.

Details and sources
attackers announced rogue BGP routes Aug 28–30 to divert Virtualizor update traffic, used a valid Let's Encrypt cert to avoid warnings, and shipped a tampered package (no cryptographic package verification on the client side) that added root SSH keys, installed Java 17 payloads, created a hidden proxyuser, and set up systemd persistence. Host AlbaHost confirmed 5 of 34 hypervisors compromised. The Hacker News
🟠 High impact

StreamRat Android banking trojan via Meta Ads

StreamRatAndroidMeta ads570,950 accounts

WhatMeta malvertising reportedly reached 570,950 accounts, pushing a near-full-control Android banking trojan to Spanish-speaking users.

Details and sources
malvertising campaign push near-full-device-control trojan to Spanish-speaking Android users; Meta ads reportedly reached 570,950 accounts. The Hacker News

Threat Actor Activity 3

🔴 Exploited or active

Pegasus zero-click spyware used against Serbian activist

NSO GroupPegasusSerbiazero-click

WhatPegasus infected a Serbian student protest member's iPhone through an iMessage zero-click exploit chain, per new forensic findings.

Details and sources
NSO Group's Pegasus infected the iPhone of a member of Serbia's student protest movement via an iMessage zero-click exploit chain, per new forensic findings. The Hacker News
🟠 High impact

RMM phishing campaign spans 46 countries, US top target

RMM phishing46 countries601 cases

WhatA large phishing operation abuses legitimate RMM tools for initial access, with 601 tracked cases and about 45% of activity aimed at the US.

Details and sources
large phishing operation abuses legitimate remote-monitoring-and-management tools for initial access; 601 tracked cases, ~45% of activity aimed at the United States. The Hacker News
🟠 High impact

Malicious .git configs can hijack AI coding agents

AI coding agents.git configs4 unpatched

WhatResearchers disclosed 8 flaws in 7 AI coding agents, including Claude, Codex and Cursor, where crafted .git configs execute attacker code; 4 remain unpatched.

Details and sources
researchers disclosed 8 flaws across 7 AI coding agents (including Claude, Codex, Cursor) where crafted .git configs can make the agent execute attacker-controlled code; 4 of the flaws remain unpatched. The Hacker News
‹ 03/09/202607/09/2026 ›