Koffei

Daily CyberNews · 11/11/2026

‹ 10/11/202612/11/2026 ›
Today in 30 seconds
  • 🔴 Patch OpenSSH on Alpine/musl now: CVE-2026-123158 SSHole is in KEV with root compromises.
  • 🔴 Update Chrome to 157.0.8103.50/.51: CVE-2026-125350 is exploited.
  • 🔴 Apply November Patch Tuesday: Outlook flaw CVE-2026-124194 is already exploited.
  • 🔴 Upgrade Xen Orchestra to 6.9.1/6.10.0: auth bypass VSA-2026-062 is exploited.
  • 🔴 Patch AhsayCBS: CVE-2026-105133 and CVE-2026-105134 were exploited as zero-days.

Vulnerabilities 14

🔴 Exploited or active

CISA KEV: OpenSSH "SSHole" + three critical pre-auth flaws

CVE-2026-123158OpenSSH SSHolein KEV

WhatThe musl-libc OpenSSH bypass is exploited with root compromises; CISA also added CVE-2026-102895, CVE-2026-116779 and CVE-2026-123218 to KEV.

DoPatch OpenSSH on musl systems plus GPUStack, FreeSWITCH and IBM Aspera Faspex; FCEB deadline is Nov 12.

Details and sources
CVE-2026-123158 ("SSHole," CVSS 8.1, EPSS ~16.5%) is an authentication-bypass bug on musl-libc Linux (e.g. default Alpine) where an oversized password makes musl's hash function return a wildcard OpenSSH treats as a match — exploitation began within ~40 hours of the fix, with confirmed root-level compromises, an actor dubbed "GT10" deploying a "TAC Agent" C2 implant, and crypto-mining/botnet use. CISA also added CVE-2026-102895 (GPUStack, CVSS 9.8), CVE-2026-116779 (SignalWire FreeSWITCH, CVSS 9.8) and CVE-2026-123218 (IBM Aspera Faspex, CVSS 9.8) to the KEV catalog the same day; all four are patched, FCEB deadline Nov 12. The Hacker News (source) · BleepingComputer (source) · OpenCVE (source)
🔴 Exploited or active

CISA KEV: Cisco BroadWorks XXE + Openwave Webtop flaw

CVE-2026-20320CVE-2026-111793in KEV

WhatCISA added a Cisco BroadWorks XXE flaw and an Openwave Webtop permissions flaw to KEV, citing active exploitation.

DoApply the existing Cisco BroadWorks and Openwave Webtop patches; FCEB deadline is Nov 24.

Details and sources
CVE-2026-20320 (Cisco BroadWorks OCI XML Parser, XXE, CVSS 7.5) lets an unauthenticated attacker read sensitive config data; CVE-2026-111793 (Openwave Messaging Webtop Media Transcoder, CVSS 4.3) lets an authenticated user process document types beyond their permissions. CISA added both to KEV Nov 10–11 citing active exploitation, despite patches having shipped one to three months earlier; FCEB deadline Nov 24. The Hacker News (source)
🔴 Exploited or active

SonicWall SonicOS SAML/MFA bypass added to KEV

CVE-2026-19868SonicOS SSL-VPNin KEV

WhatAttackers with credentials for an alternate method can skip SAML-enforced MFA on SonicOS SSL-VPN.

DoApply the SonicOS release, which also fixes CVE-2026-107155, CVE-2026-107154 and CVE-2026-82308; FCEB deadline is Nov 20.

Details and sources
CVE-2026-19868 lets an attacker with credentials for an alternate auth method skip SAML-enforced MFA on SonicOS SSL-VPN; CISA added it to KEV after reports of "possible" exploitation. SonicWall also patched CVE-2026-107155 and CVE-2026-107154 (management interface, CVSS 8.8 each) and CVE-2026-82308 (SSL-VPN pre-auth heap overflow, CVSS 7.5) in the same release. FCEB deadline Nov 20. The Hacker News (source)
🔴 Exploited or active

Chrome's 8th actively-exploited zero-day of 2026

CVE-2026-125350Chromezero-day

WhatA use-after-free in Chrome's XML component is confirmed exploited in the wild.

DoUpdate Chrome to 157.0.8103.50/.51 and update other Chromium browsers when their fixes ship.

Details and sources
CVE-2026-125350, a use-after-free in Chrome's XML component, is confirmed exploited in the wild; Google shipped an emergency fix in 157.0.8103.50/.51 (other Chromium browsers need their own updates). The Hacker News (source)
🟠 High impact

SAP November Patch Day — four unauthenticated CVSS 10.0 flaws

SAP Patch Dayfour flawsCVSS 10.0

WhatCVE-2026-120017, CVE-2026-76964, CVE-2026-89103 and CVE-2026-89126 need no authentication and allow login as any user or RCE.

DoPrioritize SAP's November patches on internet-facing SAP systems.

Details and sources
CVE-2026-120017 (SAML2 auth bypass in NetWeaver AS ABAP, login as any user), CVE-2026-76964 (ICM memory corruption/RCE), CVE-2026-89103 (SAP Gateway default-ACL broken access control) and CVE-2026-89126 (Web Dispatcher heap overflow RCE) all require no authentication or user interaction. No exploitation reported yet, but patches should be prioritized on any internet-facing SAP system. The Hacker News (source)
🟠 High impact

Fortinet discloses eight critical FortiOS/FortiManager/FortiAnalyzer flaws

CVE-2026-126170FortinetCVSS 9.8public exploits

WhatFortinet detailed eight critical FortiOS, FortiManager and FortiAnalyzer bugs, the worst an unauthenticated RCE via the JSON-RPC API.

DoApply Fortinet's October patches for FortiOS, FortiManager and FortiAnalyzer.

Details and sources
after an October "patch now, details later" rollout, Fortinet published full details Nov 10 for eight critical bugs (CVSS 9.0–9.8), worst being CVE-2026-126170 (CVSS 9.8), an unauthenticated RCE via the JSON-RPC API. No confirmed in-the-wild exploitation yet, but several have public exploit code. securityonline.info (source) · gbhackers (source)
🔴 Exploited or active

Microsoft November Patch Tuesday — 1,260 fixes, 68 critical

Patch TuesdayCVE-2026-12419468 critical

WhatMicrosoft shipped 1,260 fixes, including an already-exploited Outlook spoofing flaw, a critical LDAP RCE and Message Queuing criticals.

DoApply November Patch Tuesday, prioritizing the Outlook spoofing fix, the LDAP RCE and Message Queuing criticals.

Details and sources
part of what SANS ISC calls an "AI-induced patch surge." One flaw, CVE-2026-124194 (Outlook spoofing), is already exploited; a critical LDAP RCE and Message Queuing/Queue Manager criticals round out the priority list. SANS ISC (source)
🟠 High impact

Windows Search Component flaw among Microsoft's November fixes

Added after publishingCVE-2026-116928Windows SearchCVSS 9.8

WhatAn out-of-bounds write in Windows Search could allow unauthenticated RCE via an email attachment that Exchange indexes, possibly without opening it.

DoApply Microsoft's November fix for Windows Search.

Details and sources
CVE-2026-116928 (CVSS 9.8) is an out-of-bounds write in Windows Search that could let an unauthenticated attacker achieve remote code execution — Microsoft describes an attack delivered via a malicious email attachment that Exchange Server processes for indexing, potentially without the recipient opening it. No exploitation reported yet, but it's one of the more serious of the 48+ critical RCEs in this month's release. Malwarebytes (source)
🟡 Worth knowing

Microsoft researchers demonstrate neuron-level AI model backdooring

Microsoft researchAI model backdoorneurons

WhatMicrosoft showed editing under 10 neurons in an open-weight model plants a backdoor making agents run attacker commands on a trigger.

DoVerify checkpoints against declared base models, use trusted model registries, sandbox agents and scan for behavioral backdoors.

Details and sources
Microsoft showed that editing fewer than 10 rarely-used neurons in an open-weight model — no retraining needed — can plant a hidden backdoor that makes the model run an attacker's command whenever a chosen trigger appears in a prompt or tool output; an agent using that model then executes the command with its own permissions. Tested against Nemotron-Terminal-8B, Qwen3.6-27B and Gemma-4-31B-IT with 51–99.5% trigger success. No CVE — the technique requires the ability to modify model weights before deployment (insiders, compromised build pipelines, or tampered public model uploads). Mitigations: checkpoint verification against declared base models, trusted/private model registries, agent sandboxing and behavioral backdoor scanning. The Hacker News (source)
🟠 High impact

N-able N-central "Able2Shell" — 30+ disclosed flaws, hosted service taken offline

CVE-2026-125882Able2ShellN-able N-central

WhatA researcher published 30+ RCE and other chains in N-central, and scanning tied to the public PoC hit honeypots.

DoInstall N-central hotfix 2026.4 HF2.

Details and sources
researcher Sean Bales published a large batch of RCE/XXE/SSRF/privilege-escalation chains (7 CVEs assigned so far, e.g. CVE-2026-125882, file-repository path-traversal RCE) in the N-central RMM platform. N-able pulled its hosted service offline as a precaution and shipped hotfix 2026.4 HF2; no confirmed compromise yet, but scanning tied to the public PoC was seen against watchTowr and Defused Cyber honeypots. The Hacker News (source)
🔴 Exploited or active

Xen Orchestra auth-bypass actively exploited (VSA-2026-062)

Xen OrchestraVSA-2026-062RedLine Stealer

WhatA crafted or empty auth-token cookie lets attackers impersonate a token owner in the REST API, reaching admin and code execution on XCP-ng pools.

DoUpgrade Xen Orchestra to 6.9.1 or 6.10.0.

Details and sources
a crafted/empty auth-token cookie lets an attacker impersonate the owner of the first unexpired token in Xen Orchestra's REST API, reaching admin access and code execution on managed XCP-ng pools. No CVE yet; fixed in 6.9.1/6.10.0. Exploitation is tied to infrastructure hosting RedLine Stealer C2 and a Tor onion service. The Hacker News (source)
🟠 High impact

Git RCE via malicious recursive clone

CVE-2026-111760Gitrecursive clone

WhatA malicious repository can run code on anyone who recursively clones it by turning a submodule's .git into a symlink to attacker hooks.

DoUpgrade Git to 2.56.1 or a 2.50 to 2.55 backport.

Details and sources
CVE-2026-111760 (plus five related CVEs) lets a malicious repository achieve code execution on anyone who recursively clones it — a common CI operation — by swapping a submodule's .git into a symlink to attacker-controlled hooks. Not reported exploited yet, but high blast radius; fixed in Git 2.56.1 and backports across the 2.50–2.55 series. The Hacker News (source)
🟡 Worth knowing

Binance warns of iOS sandbox-escape PoC ("worse_query")

worse_queryiOS sandbox escapeBinance

WhatA public PoC could let a sideloaded app read other apps' private data, including wallet seed phrases; no fix or thefts confirmed.

Details and sources
a publicly released proof-of-concept could let a sideloaded app read other apps' private data, including crypto wallet seed phrases. No CVE or Apple fix confirmed yet; no confirmed thefts. The Hacker News (source)
🟡 Worth knowing

eMudhra revokes 8 more CA keys over botched key-ceremony handling

Added after publishingeMudhra8 more CAs revokeddistrust Nov 16 to 30

WhateMudhra revoked 8 more intermediate CAs over key ceremony handling, affecting 195 certificates, as major browsers move to distrust them.

Details and sources
Indian certificate authority eMudhra (emSign brand, exiting the public TLS business) revoked 8 more intermediate CAs — bringing the total to 16 after October's incident where a scripting bug truncated 8 other intermediate private keys to one byte — after finding it couldn't confirm these keys' handling met its own key-management rules during an April 2025 ceremony. No CVE, no known misuse, and no weakness demonstrated in the keys themselves; revocation reason code is "cessationOfOperation" rather than key compromise, which some researchers have questioned. 195 certificates are affected (29 eMudhra's own, the rest mostly universities and other orgs); Microsoft, Chrome, Firefox and Apple are all moving to distrust the affected roots/certificates between Nov 16 and Nov 30. The Hacker News (source)

Breaches / Incidents 9

🟡 Worth knowing

Ahold Delhaize (Albert Heijn / Etos / Gall & Gall) — attack cost quantified

Ahold DelhaizeAlbert Heijncost

WhatAhold Delhaize estimates the October attack will cut Q4 sales by €200-250 million, with no attacker named.

Details and sources
Ahold Delhaize now estimates the October cyberattack will cut Q4 sales by €200–250 million and operating profit by €75–100 million, with further remediation costs still to come. CEO Frans Muller said most disruption was contained within two weeks but full restoration of normal operations is still ongoing; no group or attacker has been named and no customer data theft has been confirmed. DutchNews (source)
🟠 High impact

Fireworks AI breach widens

Fireworks AILLMjackingthird-party cloud

WhatAttackers reached Fireworks AI's R&D environment via a third-party cloud provider, exposing billing records and partial model weights for some customers.

Details and sources
a breach first disclosed in October has grown — attackers who initially stole credentials later reached Fireworks' R&D environment via a compromised third-party cloud provider, exposing billing/invoice records, resource metadata and partial model weights/datasets for some customers. Activity is linked to a financially-motivated "LLMjacking" group; no payment card data was stored or exposed. The Record (source) · The Hacker News (source)
🟡 Worth knowing

Amplifon discloses hearing-aid customer data breach

Amplifonhearing aidsSSNs

WhatIntruders accessed Amplifon files in January and February 2026, exposing SSNs, ID numbers and health data for tens of thousands or more.

Details and sources
intruders accessed company files between Jan 18 and Feb 8, 2026 (discovered mid-February); exposed data includes Social Security numbers, driver's license/state ID numbers, and health/insurance information. Amplifon's own state filings put victims at roughly 85,804 (Texas) to 1,027 (Vermont), while plaintiffs' firms claim over 1.3 million — unconfirmed by the company. No attacker has been identified. The Record (source)
🟠 High impact

Vanderbilt University hit by cyberattack, BYOD ransomware gang claims it

Vanderbilt UniversityBYODransomware claim

WhatVanderbilt is investigating a cyberattack after the BYOD ransomware group listed it on its leak site claiming stolen data.

DoVanderbilt staff and students should disconnect devices from campus WiFi as the university instructed.

Details and sources
the university told staff and students to disconnect devices from campus WiFi as a precaution while it investigates; the ransomware group BYOD has listed Vanderbilt on its leak site claiming stolen data, alongside recent claimed attacks on other US universities. Scope not yet quantified by the university; claim unverified independently. The Record (source)
🟠 High impact

Mittwald (German hosting provider) breached via AI-assistant MCP server flaw

MittwaldMCP server56 projects

WhatAttackers exploited a flaw in Mittwald's MCP server, compromising 56 customer projects and going undetected for eight days.

Details and sources
attackers exploited a flaw in Mittwald's Model Context Protocol server — used to connect customer AI assistants to its hosting-management platform — compromising 56 customer projects and going undetected for eight days. Mittwald's own characterization of the intrusion as "AI-assisted" rests only on unusually fast lateral movement in its logs, not independent confirmation. No further attacker activity detected since; forensic review ongoing. The Record (source)
🟠 High impact

NEAR AI breached via SGLang inference-engine RCE

NEAR AISGLangRCE

WhatAn SGLang RCE let an attacker run code on a NEAR AI inference server, exposing about 100 users' data and two customers' API keys.

Details and sources
an attacker exploited a remote-code-execution flaw in the open-source SGLang inference engine to run code on an inference server, exposing account data for roughly 100 users and chat logs/credentials for 18 legacy AI agents, plus two customers' API keys (since revoked). The flaw even allowed access to requests meant to be isolated inside a trusted execution environment. The Hacker News (source)
🟡 Worth knowing

Binance denies alleged "ShinyHunters" breach claim (335 million users)

BinanceShinyHuntersclaim denied

WhatA seller posing as ShinyHunters offered data on 335 million Binance users, but Binance says it found no breach.

Details and sources
a threat actor posing as ShinyHunters listed data for 335 million Binance users for sale on a new cybercrime forum, demanding $900,000. Binance says it found no evidence of a breach and that the data resembles past scam listings sourced from infostealer malware on users' own devices rather than a compromise of its systems. BleepingComputer (source)
🟠 High impact

Russian hosting giant Beget hit by destructive wiper attack

BegetRussiawiper attack

WhatAn unattributed attacker reached Beget's infrastructure management layer and ran data-zeroing scripts across VPS, S3 and Kubernetes systems.

Details and sources
an unattributed attacker reached the management layer of Beget's entire infrastructure on Nov 1 and ran data-zeroing scripts across VPS, S3 and Kubernetes systems — pure destruction, no evidence of data theft or a ransom demand. As of Nov 9, 60% of shared hosting and 40% of dedicated servers were restored, with roughly 1,000 servers still needing manual security checks. BleepingComputer (source)
🟠 High impact

Railway says attacker stole customers' AI provider keys

RailwayAI provider keysbuild system

WhatAn attacker moved through Railway's build system to decrypt customer environment variables and steal AI provider keys, affecting up to ~100,000 workspaces.

Details and sources
in October 2026 an attacker moved through Railway's build system and internal tools to reach a machine able to decrypt customer environment variables, then systematically searched for AI provider API keys; also exposed were linked-account tokens for 1,852 users (including GitHub OAuth tokens), some cloud/database/payment credentials matched by the attacker's queries, and stored histories of 353 builds. Up to ~100,000 workspaces (1.85%) may have been affected; Railway estimates roughly $740,000 in unauthorized AI spending and says it will cover charges providers won't refund. No customer databases or code repositories were accessed; the attacker was removed Oct 21, and Railway has filed a complaint with the FBI's IC3. The Hacker News (source)

Malware / Ransomware 3

🔴 Exploited or active

AhsayCBS backup-server zero-days chained to deploy backdoor and miners

CVE-2026-105133CVE-2026-105134AhsayCBS

WhatAttackers chained two AhsayCBS zero-days to bypass authentication, dropping web shells, miners and a Python backdoor, with interest in an adjacent Veeam server.

DoApply Ahsay's mid-October AhsayCBS patches and check backup consoles for web shells, XMRig and tunneling or RMM tools.

Details and sources
attackers chained CVE-2026-105133 and CVE-2026-105134 as zero-days (patched mid-October) to bypass authentication on Ahsay's AhsayCBS backup console, initially dropping JSP web shells and XMRig miners, then a more patient intrusion installing a custom sandbox-evading Python backdoor and tunneling/RMM tools — with signs of interest in credentials on an adjacent Veeam backup server. Backup infrastructure is a classic ransomware pre-positioning target. The Hacker News (source)
🟠 High impact

PapRAT — new RAT from Bitter APT targets South Asia

BitterPapRATSouth Asia

WhatBitter added a new C++ and Rust RAT with command execution, file operations and screenshots, delivered via LNK, VHDX and WebDAV chains.

Details and sources
Bitter (APT-C-08) added a previously unreported RAT, written in both C++ and Rust, to its South Asia-focused espionage toolkit, delivered via LNK, VHDX and WebDAV infection chains alongside its known WmRAT malware. Capabilities include command execution, file operations, screenshots and (in the Rust variant) PowerShell execution. The Hacker News (source)
🟠 High impact

ARTEX AI-agent framework repurposed for autonomous criminal campaigns

ARTEXAI agentscriminal campaigns

WhatCheck Point documented ARTEX autonomously running campaigns that stole 300+ iOS signing certificates and 190,000+ gaming credentials.

Details and sources
Check Point documented real-world criminal use of ARTEX, an AI-agent orchestration framework originally built for pen-testing, to autonomously run three separate campaigns — theft of 300+ iOS app-signing certificates, harvesting of 190,000+ gaming credentials across 13 marketplaces, and exploitation of 292 cryptocurrency-related domains — with operators giving agents only a high-level objective and Chinese LLMs (notably DeepSeek) doing the rest. eSecurity Planet (source)

Threat Actor Activity 6

🟠 High impact

France formally blames Iran's IRGC for the October supermarket refrigeration sabotage

Shahid KavehCyberAv3ngersIRGC

WhatFrance formally attributed October's refrigeration sabotage at about 700 retail sites to an IRGC unit and will pursue EU sanctions and arrest warrants.

Details and sources
France's Cyber Crisis Coordination Centre (ANSSI, military cyber command, DGSE, DGSI) attributed October's refrigeration-controller sabotage at roughly 700 French retail sites — which caused tens of millions of euros in food losses and at least 15 fires — to Shahid Kaveh, an IRGC cyber-offensive unit the US has linked to CyberAv3ngers. Foreign Minister Jean-Noël Barrot confirmed IRGC responsibility at a G7 meeting, France summoned Iran's ambassador, said it will pursue EU sanctions, and the Paris Prosecutor's Office opened a judicial investigation seeking international arrest warrants for the unit's leadership. Attribution is based on shared reconnaissance infrastructure, TTP overlap, Farsi-language traces and a linked propaganda outlet — officially confirmed, though not courtroom proof. The Record (source) · The Hacker News (source)
🟠 High impact

Turla deploys new "Cereal" toolset against NATO/EU-linked government mail

TurlaCerealExchange C2

WhatTurla's new Cereal toolset turns compromised Exchange servers into covert C2 to harvest NATO and EU-related mail in Kazakhstan and Romania.

Follow-ups: 12/11
Details and sources
ESET documented a new, modular Turla (Secret Blizzard) toolset called Cereal that turns a compromised Microsoft Exchange server into its own covert C2 channel — avoiding blockable external infrastructure — used against government targets in Kazakhstan and Romania, with the Romanian implant specifically configured to harvest NATO/EU-related correspondence. The Hacker News (source) · The Record (source)
🔴 Exploited or active

Blind Eagle runs fake Colombian court-notice phishing campaign across Latin America

Blind EagleLatin Americafake court notices

WhatBlind Eagle is phishing Latin American judicial, administrative and educational organizations with fake Colombian court notices that deploy a new AsyncRAT/DCRat-derived RAT.

Details and sources
Blind Eagle (APT-C-36), via a Brazilian-linked operational cluster, is targeting judicial, administrative and educational organizations across Latin America with phishing emails disguised as Colombian prosecutor/court notices, using HTML smuggling and AMSI/ETW-evading VBScript to deploy a new AsyncRAT/DCRat-derived RAT. No confirmed successful victims reported yet. The Hacker News (source)
🟡 Worth knowing

Check Point links three China-nexus espionage campaigns via shared "Sparrow" toolkit

CopperDragonFamousSparrowSalt Typhoon

WhatCheck Point ties three China-nexus espionage campaigns to a shared Sparrow toolkit developed by CopperDragon.

Details and sources
Check Point Research ties together SilentDelivery (fake-certificate watering holes across Europe/Asia), OceanGate (Linux-to-Windows lateral movement in Asia) and DarkWire (Latin American telecom/government targeting) through a common toolset it calls "Sparrow," attributing tool development to an entity dubbed CopperDragon that appears to supply multiple Chinese APT clusters including FamousSparrow, Earth Estries and Tropic Trooper, with a possible (unconfirmed) overlap with Salt Typhoon. The Hacker News (source)
🟠 High impact

NoName057(16) doxxes European officials in apparent retaliation

NoName057(16)doxxingEuropean officials

WhatA site linked to NoName057(16) published personal data on about 18 European officials, prosecutors and journalists.

Details and sources
a dark-web site linked to pro-Russian hacktivist group NoName057(16) published personal data on roughly 18 European officials, prosecutors and journalists — including Italian Foreign Minister Antonio Tajani and Germany's BKA president — in what researchers assess as retaliation for a 2025 law-enforcement operation against the group. Russian state direction is unconfirmed; Russia denies involvement. The Record (source)
🟡 Worth knowing

Ukraine arrests four behind fake investment call-center fraud network

Ukrainearrestsinvestment fraud

WhatUkrainian police arrested four alleged leaders of a fake investment call center that defrauded 135+ German victims of about €4.3 million.

Details and sources
Ukrainian police arrested four alleged leaders of a fraud operation that posed as financial advisers to defraud at least 135 German victims of roughly €4.3 million via fake crypto/stock platforms, laundering proceeds through accounts and crypto wallets. The Record (source)
‹ 10/11/202612/11/2026 ›