Koffei

Daily CyberNews · 10/11/2026

‹ 09/11/202611/11/2026 ›
Today in 30 seconds
  • 🔴 Update Chrome to 157.0.8103.50/.51: CVE-2026-125350 is exploited in the wild.
  • 🔴 Patch Xen Orchestra to 6.9.1/6.10.0: auth bypass VSA-2026-062 is actively exploited.
  • 🔴 Apply the SonicOS fix for CVE-2026-19868 MFA bypass, in KEV with a Nov 20 deadline.
  • 🔴 Hunt for poisoned bee-js 13.3.0 and Swarm Desktop 0.56.2, which steal wallet keys.
  • 🟠 Patch SAP now: four CVSS 10.0 flaws, two with no workaround.

Vulnerabilities 11

🔴 Exploited or active

SonicWall SonicOS SSL-VPN SAML/MFA bypass actively exploited

CVE-2026-19868SonicOS SSL-VPNCVSS 7.5in KEV

WhatAttackers with credentials for one method can bypass SAML-enforced MFA on Gen7/Gen8 firewalls up to 7.3.3-7015 or 8.2.2-8015.

DoApply SonicWall's published fix; federal agencies must remediate by November 20.

Details and sources
CVE-2026-19868 is an authentication-enforcement flaw (CWE-308) in SonicOS's SSLVPN service that lets an attacker holding valid credentials for one authentication method bypass SAML-enforced checks, including SAML-enforced MFA, on Gen7/Gen8 firewalls running 7.3.3-7015 or 8.2.2-8015 and earlier. CVSS 7.5. SonicWall's PSIRT says it is investigating a report of potential exploitation, and CISA added the bug to its KEV catalog on November 6, 2026 with a federal remediation deadline of November 20; France's cyberveille health-security portal also lists it as actively exploited with a fix already published. SecurityAffairs (source) · BleepingComputer (source)
🟡 Worth knowing

Apple patches two iMessage group-chat flaws

Added after publishingCVE-2026-113961CVE-2026-113960iMessage

WhatTwo iMessage group chat flaws let an attacker join a chat and receive messages, or read messages sent before joining.

DoUpdate iPhones and iPads to iOS/iPadOS 27.2.1.

Details and sources
CVE-2026-113961 let a remote attacker join a group chat and receive its messages via an authorization flaw; CVE-2026-113960 let someone added to a group read messages sent before they joined. Both fixed in iOS/iPadOS 27.2.1; Apple's notes don't say either was exploited, and some older devices (iPhone XS/XS Max/XR, 7th-gen iPad, some Apple Watch models) don't list the fix. The Hacker News (source)
🟡 Worth knowing

Binance warns of iOS sandbox-escape PoC risking crypto wallets

Added after publishingworse_queryiOS sandbox escapeBinance

WhatA public PoC lets a sideloaded app read other apps' containers on iOS 26/27, which could expose seed phrases and keys.

DoRemove sideloaded or untrusted apps, update iOS, and move funds to a new wallet on a clean device if compromise is suspected.

Details and sources
a publicly released proof-of-concept called "worse_query" lets a sideloaded app read other apps' file containers on iOS 26/27, which Binance says could expose seed phrases, private keys, and login credentials. No CVE assigned and no confirmed attacks yet; it requires sideloading (AltStore/SideStore), not an App Store install, and the researcher released it early after saying Apple was slow to respond. Apple's iOS 26.7.3/27.2.1 updates don't mention it. Binance advises removing sideloaded/untrusted apps, updating iOS, and — if compromise is suspected — moving funds to a new wallet on a clean device. The Hacker News (source)
🔴 Exploited or active

Xen Orchestra authentication bypass actively exploited

Added after publishingXen OrchestraVSA-2026-062exploited

WhatA crafted authenticationToken cookie logs attackers in as the first unexpired token's owner, giving Admin rights and root takeover of XCP-ng nodes.

DoUpgrade to Xen Orchestra 6.9.1/6.10.0 or xo-server 5.211.2/5.213.1; otherwise remove internet exposure and block the empty-token pattern at a WAF.

Details and sources
tracked by Vates VMS as VSA-2026-062 (no CVE yet), a crafted cookie (authenticationToken=j:%7B%7D) is parsed as an empty object and passed to the token provider unchecked, which then returns the first unexpired token and authenticates the attacker as its owner — giving Admin privileges, code execution on managed pools, and root takeover of an XCP-ng node. Vates VMS reports active exploitation, linked to an IP flagged as a RedLine Stealer C2 server plus a Tor domain, though no formal actor attribution yet. Fixed in Xen Orchestra 6.9.1/6.10.0 and xo-server 5.211.2/5.213.1; where patching isn't immediate, avoid internet exposure, block the known IP/domain and the empty-token request pattern at a WAF, and rotate credentials if compromise is suspected. Discovered by Filip Magic and Christoffer Olsen of TDC NET Security. The Hacker News (source)
🟠 High impact

Git patches flaw letting a malicious repo execute code via submodule symlink trick

Added after publishingCVE-2026-111760Gitsubmodule symlink

WhatAn integer truncation lets a malicious repo swap a submodule's .git for a symlink to malicious hooks that run on recursive clone.

DoUpgrade Git to 2.56.1 or a 2.50.2 to 2.55.1 backport, and avoid recursive clones of untrusted repositories until patched.

Details and sources
CVE-2026-111760 is an integer truncation in Git's index handling — an overly long path slips past Git's safety check and gets written to disk under a shortened name. An attacker can arrange for that shortened name to be .git, letting them swap a freshly-cloned submodule's .git file for a symlink to a folder holding malicious hooks, which then execute on recursive clone. Fixed in Git 2.56.1 and backports 2.50.2–2.55.1; GitHub has applied the fix to github.com, GitHub Enterprise Cloud, and Enterprise Server, though Git for Windows had not yet shipped a 2.56.1-based build as of the report. Until patched, avoid recursive clones of untrusted repositories. The release also fixes five other CVEs. The Hacker News (source)
🟠 High impact

SAP's November Patch Day fixes four maximum-severity (CVSS 10.0) flaws

Added after publishingSAP Patch Dayfour flawsCVSS 10.0

WhatCVE-2026-76964 and CVE-2026-89126 allow unauthenticated RCE, CVE-2026-89103 enables Gateway impersonation, and CVE-2026-120017 lets attackers log in as any user.

DoApply SAP's November patches; disabling WebSocket support temporarily mitigates the two memory bugs, but the Gateway and SAML2 flaws have no workaround.

Details and sources
CVE-2026-76964 and CVE-2026-89126 are heap out-of-bounds-write memory corruption bugs (malformed WebSocket frames) in NetWeaver AS ABAP's ICM and in SAP Web Dispatcher respectively, both unauthenticated and network-reachable, leading to code execution. CVE-2026-89103 is an SAP Gateway access-control flaw where default rules trust external callers as the system's own application servers, enabling impersonation. CVE-2026-120017 is a SAML 2.0 authentication bypass in NetWeaver AS ABAP/ABAP Platform where assertion signatures aren't checked in some configurations, letting an attacker log in as any user including privileged ones. No exploitation reported yet; no workaround exists for the Gateway or SAML2 flaws, while disabling WebSocket support temporarily mitigates the other two at a feature cost. The Hacker News (source)
🟠 High impact

Barracuda Email Security Gateway auth-bypass-to-root chain

Added after publishingCVE-2026-116606CVE-2026-111560Barracuda ESG

WhatAn ESG login authentication bypass chained with a timezone command injection gives full root takeover across six ESG product lines.

DoPatch Barracuda ESG, since there is no workaround.

Details and sources
CVE-2026-116606 (CVSS v4.0 9.3, Critical; CWE-288) is an authentication bypass in ESG's login handling — Barracuda says exploitation requires some prior access, but Mandiant's independent analysis found it can be triggered pre-authentication. Chained with it, CVE-2026-111560 (CVSS v4.0 8.6, High; CWE-78) is an OS command-injection flaw in the appliance's timezone-configuration handling that then executes as root, giving an authenticated-or-not attacker full appliance takeover. No in-the-wild exploitation confirmed yet, but the chain affects six Barracuda ESG product lines and has no workaround beyond patching. SecurityWeek (source)
🔴 Exploited or active

Openwave Webtop Media Transcoder flaw sees fresh exploitation report

Added after publishingCVE-2026-111793Openwave WebtopCVSS 4.3in KEV

WhatAn authenticated user can send blocked file types to the Webtop transcoder to expose sensitive data; exploitation is now reported and it entered KEV.

DoUpgrade Webtop to 1.1.22.0 or later, or move to the Media Server 2.0 line.

Details and sources
CVE-2026-111793 (CWE-669, CVSS 4.3 Medium, EPSS 0.27%) lets an authenticated user who bypasses the webmail front-end send blocked file types directly to the Webtop transcoder, which JPCERT/CC says can expose sensitive information from a specially crafted file. CISA's assessment as of October 19 had listed exploitation as "none," but JPCERT/CC's advisory now cites the developer's own statement that exploitation has been observed, with OpenCVE recording a KEV-catalog addition on November 10 (due November 24). No specific campaign or threat actor is named. Fixed since version 1.1.22.0 (released December 2021) — affected operators should upgrade or move to the current Media Server 2.0 line. SecurityOnline (source)
⚪ Context

Cisco BroadWorks XXE flaw resurfaces — an older disclosure, not new news

Added after publishingCVE-2026-20320Cisco BroadWorksCVSS 7.5

WhatAn XXE flaw in BroadWorks patched in August resurfaced in CVE records, with no exploitation and no new development.

Details and sources
CVE-2026-20320 (CVSS 7.5 High, CWE-611) is an unauthenticated, out-of-band XML external entity injection flaw in BroadWorks that lets a remote attacker read sensitive files on the host. This was originally disclosed and patched via Cisco's advisory on August 19, 2026; it is not in CISA's KEV catalog and there is no evidence of exploitation. It is included here only because its CVE record was recently revisited — no new development to report. Cisco Security Advisory (source)
🟡 Worth knowing

Microsoft patches Outlook spoofing flaw on Patch Tuesday

Added after publishingCVE-2026-124194OutlookCVSS 6.5

WhatA spoofing flaw in Outlook could let a crafted message mislead a user about its sender; no exploitation reported.

DoApply this month's Outlook updates.

Details and sources
CVE-2026-124194 (CVSS 6.5 Medium, CWE-356) is a spoofing vulnerability in Microsoft Outlook that could let a crafted message mislead a user about its sender or origin. Not listed in CISA's KEV catalog and no exploitation reported; a routine fix to apply with this month's other Outlook updates. OpenCVE (source)
🟠 High impact

eMudhra CA key ceremony flaw exposed eight recoverable intermediate certificate private keys

Added after publishingeMudhraCA private keysemSign ECC Root CA - G3

WhatEight eMudhra intermediate CA keys had weak randomness and could be recomputed from public certificates, prompting revocation and browser distrust of the root.

Details and sources
a researcher reported via the Chrome Root Program on October 17 that eight of eMudhra's intermediate CA private keys — generated in an April 2025 key ceremony — had insufficient randomness, letting anyone recompute the private key from the public certificate in under a second; one researcher recovered all eight that way. Six are TLS intermediates (DV/OV/EV, each in G3B/G3C variants) chaining to the emSign ECC Root CA - G3 that Firefox and Chrome trust, which would have let a holder issue valid certificates for any domain; the other two are code-signing intermediates trusted by Microsoft. eMudhra revoked all eight on October 18, and within days Chrome, Firefox, Apple and Microsoft blocked the certificates and Mozilla and Google moved to distrust the root entirely; Go, Node.js, Python's certifi, Rust's webpki-roots and several Linux distros shipped matching fixes. eMudhra says only one of the eight was actually deployed in production and none had issued end-entity certificates, and no confirmed malicious use has been found — though detection is limited to what's visible in Certificate Transparency logs. Mozilla's bug has since widened to cover related CRL-signature problems and questions about eMudhra's transparency on an earlier incident, with eMudhra still owing the Root Program a full root-cause report. Mozilla Bugzilla (source) · The Hacker News (source)

Breaches / Incidents 11

🟠 High impact

Iris Ohyama cloud breach exposes up to 7.7 million

Iris OhyamaJapan7.7 million

WhatA suspicious file in a third-party cloud behind the Iris Plaza store exposed names, addresses, birth dates and encrypted passwords of up to 7.7 million people.

Details and sources
the Japanese household-goods maker found a suspicious file in a third-party cloud service behind its Iris Plaza online store, first disclosed October 29 and now estimated at up to 5.8 million registered/former members plus 1.9 million non-members. Exposed data includes names, addresses, phone numbers, emails, gender, dates of birth and encrypted account passwords; no credit card data was stored there. No attacker contact or ransom demand has surfaced, and as of November 5 the company had no reports of misuse. BleepingComputer (source)
🟡 Worth knowing

Amplifon hearing-health breach notices reach state regulators

Amplifonhearing healthSSNs

WhatAmplifon reported a breach exposing Social Security numbers and health records to US state regulators, without a full timeline or total count.

Details and sources
Amplifon (USA) Inc. disclosed a breach to the Vermont Attorney General on November 5, reporting 1,027 affected residents, with notices also filed in Texas and roughly a dozen other states; exposed data includes Social Security numbers and health records, though the company has not yet disclosed the full breach timeline or total number affected. claimdepot.com (source)
🟡 Worth knowing

Albert Heijn/Ahold Delhaize cyberattack now put at up to €250m in lost Q4 sales

Ahold DelhaizeAlbert Heijn€250m

WhatAhold Delhaize said the October attack will cut Q4 sales by €200-250 million and operating profit by €75-100 million.

Details and sources
updating the October 11 Netherlands incident, Ahold Delhaize says the attack will cut Q4 sales by €200-250 million and operating profit by €75-100 million, on top of ongoing remediation costs. CEO Frans Muller said most disruption was contained within two weeks, but full normal operations are still being restored. No attacker attribution or confirmed data theft has been disclosed for this incident. DutchNews (source)
🔴 Exploited or active

Second wave of European supermarket refrigeration attacks hits 100+ stores

DanfossCopeland100+ storesEurope

WhatRefrigeration units overheated overnight Nov 1 to 2 at 100+ stores in six countries, via internet-exposed and often end-of-life controllers.

Details and sources
weeks after the October 11-12 attack on ~700 French stores, refrigeration and freezer units overheated overnight November 1-2 at more than 100 stores across Belgium, Estonia, Ireland, Luxembourg, the Netherlands and Sweden, with problems also reported in Finland, Italy and Portugal. Retailers named include Delhaize, Carrefour, Colruyt/Spar, SuperValu, ICA, Coop and Selver; affected equipment again appears to be internet-exposed Danfoss and Copeland controllers, many end-of-life with no patch available. No group has claimed responsibility and investigators have not confirmed a link between the two waves; Belgium's federal prosecutor has opened an investigation. SecurityWeek (source) · ACHR News (source)
🟠 High impact

N-able takes Hosted N-central offline after 30+ zero-days disclosed

Able2ShellN-able N-central30+ flaws

WhatA researcher released PoCs for over 30 N-central flaws including root RCE chains, and N-able took Hosted N-central offline.

DoOn-premises N-central customers should shut down their servers until N-able ships a hotfix.

Details and sources
independent researcher Sean Bales publicly released proof-of-concept exploits for a set of more than 30 vulnerabilities in N-able's N-central RMM platform, collectively dubbed Able2Shell — including credential disclosure, arbitrary file writes, directory traversal, XXE injection, and chains leading to root-level remote code execution. N-able took its Hosted N-central environment offline as a precaution and told on-premises customers to shut down their servers pending a hotfix; it says it has not seen signs of active exploitation. The Hacker News (source)
🟡 Worth knowing

Researcher finds 100+ flaws across UK government code repos, including "god-tier" credential exposure

UK government reposPlanning InspectorateEntra ID

WhatA researcher using Claude found 100+ serious flaws in UK government repos, including exposed Entra ID credentials for the Planning Inspectorate.

Details and sources
security researcher Chris Nesbitt-Smith, using consumer-grade Claude models, found over 100 serious vulnerabilities across roughly two-thirds of 349 public UK government code repositories spanning 33 public-sector organizations. The worst let anyone extract Microsoft Entra ID credentials for the Planning Inspectorate, which would have given read-write access to all its SharePoint sites, private staff OneDrive accounts, and sensitive planning-evidence libraries; the Inspectorate fixed the exposed credentials within 20 minutes of the report and found no evidence of a prior breach. WIRED (source)
🔴 Exploited or active

Chrome patches actively-exploited zero-day

CVE-2026-125350Chromezero-day

WhatA use-after-free in Chrome's XML component has a known exploit in the wild, the eighth exploited Chrome zero-day of 2026.

DoUpdate Chrome to 157.0.8103.50/.51 via the About page and relaunch.

Details and sources
Google shipped an emergency Chrome update fixing CVE-2026-125350, a use-after-free in Chrome's XML component that Google says has a known exploit in the wild — the eighth actively-exploited Chrome zero-day of 2026. Fixed in 157.0.8103.50/.51 (Windows/macOS) and 157.0.8103.50 (Linux); users should update via Chrome's About page and relaunch. The Hacker News (source) · Malwarebytes (source)
🟡 Worth knowing

Microsoft Paint RCE patch mix-up leaves fixed version unclear

CVE-2026-83844Microsoft PaintCVSS 8.1

WhatMicrosoft listed conflicting fixed versions for the Paint RCE, and the Store reportedly still offers the vulnerable build.

DoCheck the installed Paint version directly instead of trusting an up to date status.

Details and sources
CVE-2026-83844 (CVSS 8.1) is a zero-click, out-of-bounds-read flaw in Windows Paint reachable over the network without user interaction. Microsoft's advisory has listed conflicting fixed-version numbers (11.2605.91.0, then 11.2609.401.0), and admins report the Microsoft Store still offering the older, vulnerable build — administrators are advised to check the installed Paint version directly rather than trust an "up to date" status. Dark Reading (source)
🟡 Worth knowing

Claude-related Linux code-execution flaw reported, dubbed "Office Party"

Office PartyClaudeLinux

WhatDark Reading reports attackers abusing a flaw called Office Party to run code on Linux via Claude; details are unconfirmed.

Details and sources
Dark Reading reports attackers are abusing a flaw it calls "Office Party" to run malicious code on Linux systems via Claude. The publication's page could not be fetched to confirm further technical detail (affected component, discoverer, patch status), so treat the specifics as unconfirmed pending a readable source. Dark Reading (source)
🟠 High impact

France formally attributes the October refrigeration-sabotage attacks to Iran's IRGC

Added after publishingShahid KavehIRGCFrance

WhatFrance attributed the October sabotage of refrigeration controllers at about 700 retail sites, which caused food losses and fires, to an IRGC unit.

Details and sources
the French government attributes the October 11 attack that hit remote refrigeration controllers at roughly 700 retail sites (Casino, Leclerc, Coopérative U, Metro, Carrefour, Intermarché, and smaller food businesses), causing food losses and about 20 fires, to Shahid Kaveh, an IRGC unit that has claimed attacks under hacktivist personas like CyberAveng3rs and Cyber Isnaad Front. Cited evidence includes shared attack infrastructure also used to scan Schneider Electric gear at French hydroelectric sites, use of the FOFA scanning tool, Persian-language artifacts and Iran-based IPs, and a claim posted by an outlet linked to the unit on October 13. France's Foreign Ministry calls it the first purely-sabotage offensive cyber campaign on French soil, plans to summon Iran's ambassador and seek EU sanctions, and the Paris prosecutor has opened an investigation and requested international arrest warrants for the unit's leadership. The attribution does not cover the separate second wave of refrigeration attacks that hit other European countries in early November. The Hacker News (source)
🟠 High impact

NEAR AI breached via RCE flaw in its SGLang inference engine

Added after publishingNEAR AISGLangRCE

WhatAn attacker exploited an unpatched SGLang RCE on a NEAR AI inference server, reaching account data for about 100 users and two Cloud API keys.

Details and sources
an attacker exploited an unpatched remote code execution flaw in SGLang (an open-source inference engine running inside NEAR AI's trusted execution environment) to run code on one of NEAR AI's inference servers, potentially exposing that day's processed requests including confidential-mode ones. From there the attacker reached an internal system behind the deprecated Private Chat product, downloading account details (emails, names, and for some accounts plan/balance info) for about 100 users, plus chat history, settings and logs from 18 AI agents on an older, non-TEE NEAR AI Agents deployment; two customers' Cloud API keys were exposed in logs and have been revoked. No evidence stored conversations or uploaded files were accessed. NEAR AI patched the SGLang flaw on all its inference servers, revoked active sessions/admin credentials, took Private Chat offline, and has restored its Cloud service; no CVE or affected-version range has been published yet. The Hacker News (source)

Malware / Ransomware 3

🔴 Exploited or active

Supply-chain attack on Ethersphere's Swarm network via poisoned npm package

Etherspherebee-js 13.3.0Swarm Desktop 0.56.2

WhatAttackers with stolen CI credentials published a malicious bee-js release and trojanized Swarm Desktop that steal node wallet keys and install remote shells.

DoHunt for bee-js 13.3.0 and Swarm Desktop 0.56.2 installs, the evjawbreaker shell, or unexpected ScreenConnect clients.

Details and sources
attackers used stolen CI credentials to publish a malicious bee-js 13.3.0 release and a trojanized Swarm Desktop 0.56.2 build with valid code-signing, both distributed through Ethersphere's own legitimate release pipeline. Both drop a loader that steals Swarm node wallet keys and installs a Go-based remote shell ("evjawbreaker," controlled via an Ethereum smart-contract address) on Linux/macOS or a genuine, abused ScreenConnect client on Windows. Socket and Pluto Security link the payloads with high confidence to the group behind 2025's SleepyDuck Open VSX extension attack; Ethersphere has pulled the malicious versions and tightened its signing-pipeline approvals. The Hacker News (source)
⚪ Context

AI tools said to be making reverse-engineering of software and malware far easier

Added after publishingAIreverse engineeringDark Reading

WhatDark Reading reports AI tools are making reverse engineering of software and malware easier; details are unconfirmed.

Details and sources
Dark Reading reports that AI-assisted reverse-engineering tooling is lowering the skill bar for both security researchers and malware authors analyzing or cloning mainstream software and malware. The page could not be fetched to confirm further technical detail (specific tools, named researchers, concrete examples), so treat the specifics as unconfirmed pending a readable source. Dark Reading (source)
🔴 Exploited or active

Acronis finds 575+ malicious AI-agent "skills" on the OpenClaw marketplace

OpenClaw575+ malicious skillsAcronis

WhatOver 575 malicious AI-agent skills on the OpenClaw marketplace deliver trojans, miners and infostealers to Windows and macOS users.

Details and sources
Acronis's Threat Research Unit reports over 575 malicious skills distributed through the OpenClaw AI-agent marketplace, delivering trojans, cryptominers and infostealers to Windows and macOS users who install them. The same report documents a separate AI-orchestrated attack operation (see BOSSAI below) spanning Taiwan, Thailand and Mexico. Help Net Security (source)

Threat Actor Activity 7

🟠 High impact

Check Point links three China-nexus espionage campaigns to a shared toolmaker, "CopperDragon"

Added after publishingCopperDragonFamousSparrowSparrow framework

WhatCheck Point links the SilentDelivery, OceanGate and DarkWire espionage campaigns to CopperDragon, a toolmaker supplying several China-nexus groups.

Details and sources
Check Point Research assesses CopperDragon as the developer behind the Sparrow malware framework, supplying tools to China-nexus groups including FamousSparrow, Earth Estries, Tropic Trooper, UAT-9244 and UAT-8302 (with FamousSparrow assessed to overlap somewhat with Salt Typhoon). SilentDelivery (since early 2026) hit government, shipping, IT and research targets across Europe and Asia (Czech Republic, Germany, France, Taiwan, Nepal, Hong Kong, Indonesia) via watering-hole sites pushing a trojanized "Certificate Fix" MSI, plus phishing with DLL-sideloading LNK files, delivering SquadC/SquawkDoor or a SparrowDoor variant. OceanGate (2025–2026) targeted mostly Asian organizations, pivoting from internet-facing Linux servers to Windows via RDP password-guessing, using WinDivert to hide a SparrowDoor backdoor behind port 443, and deploying new Linux backdoors AnnobinDoor/SparrowShell alongside Windows tools showing possible code/infrastructure overlap with Mélofée. DarkWire (mid-2025) hit telecom and government targets in Latin America via DLL sideloading and locally-based infrastructure, using CopperDoor/SparroWocky, DNS-tunneling CopperDNS, and LSASS-dumping CopperLsass. The Hacker News (source)
🟠 High impact

Turla deploys new "Cereal" framework abusing Exchange servers as C2

Added after publishingTurlaCerealExchange C2

WhatTurla used a new Cereal toolset running inside compromised Exchange servers as its own C2 against government targets in Kazakhstan and Romania.

Follow-ups: 12/11
Details and sources
ESET documents a previously unknown toolset, Cereal, used by Russia-aligned Turla (aka Snake) against government targets in Kazakhstan and Romania in late 2025 — in the Romanian case configured to harvest emails tied to NATO and EU institutional addresses. It runs entirely inside compromised Microsoft Exchange servers (injecting into EdgeTransport.exe), using them as its own C2 so there's no external infrastructure to spot; commands arrive as in-memory .NET programs. Modules include CerealRunner (deployment), CerealNeuron (a passive backdoor disguised as normal Exchange traffic), CerealHermit, and CerealBox (the email harvester, which filters, compresses, encrypts and exfiltrates matching DOC/DOCX/PDF attachments). Initial access vector is unknown. The Hacker News (source)
⚪ Context

Network of suspicious advent-calendar storefronts flagged ahead of the holidays

Added after publishingMalwarebytesadvent calendar storefrontsscams

WhatMalwarebytes flagged related stores selling unlicensed advent calendars with misleading sales tactics, but found no payment card theft.

Details and sources
Malwarebytes researchers identify a cluster of related online stores selling unlicensed Zelda-, Michael Jackson-, Lord of the Rings-, Harry Potter- and Metallica-branded advent calendars, sharing a common "sales playbook": fake countdown timers and stock warnings, manually-entered "verified buyer" reviews, contradictory product claims, copied storefront templates, AI-generated product photos, unsubstantiated "Walmart partnership" claims on TikTok, no listed business address/phone, and costly optional add-ons like "shipping protection." No evidence was found that the sites steal payment card data — the risk identified is misleading sales claims and orders that arrive wrong or not at all. Malwarebytes (source)
🟠 High impact

Bitter APT deploys new PapRAT backdoor against South Asian targets

BitterPapRATSouth Asia

WhatFortiGuard attributes the new PapRAT backdoor, in C++ and Rust builds, to Bitter, delivered via LNK, VHDX and WebDAV chains.

Details and sources
FortiGuard Labs (researcher Cara Lin) attributes a previously unreported remote access trojan, PapRAT (with C++ and Rust builds), to Bitter (aka APT-C-08/Hazy Tiger/T-APT-17/TA397). It arrives via three infection chains — a malicious LNK abusing ftp.exe, a VHDX disk image hiding an MSI installer behind a lure spreadsheet, and a WebDAV-delivered LNK — with one chain also dropping Bitter's known WmRAT. The Hacker News (source)
🟡 Worth knowing

Single Taiwan-based hacker ran a 9-agent AI "attack mesh" across 11 countries

BOSSAITaiwanAI attack mesh

WhatAcronis attributes a campaign across 11 countries to one Taiwan-based hacker using uncensored local AI models.

Details and sources
Acronis's Threat Research Unit attributes a wide-ranging intrusion campaign — branded BOSSAI after the attacker's own tooling — to one individual in Hualien City, Taiwan, who used uncensored local AI models (abliterated DeepSeek, GLM and Qwen builds) after commercial models refused his attack requests. Confirmed access/data collection hit Taipei's 119 emergency-dispatch API, a Taipei fire-department patient platform, a Taiwanese lending firm, three Thai universities and roughly 206,800 Mexican phone records; the attacker was identified after his own poorly secured server, holding 90,000+ files of logs and credentials, revealed a reused identity and his home IP address. The Hacker News (source)
🟠 High impact

Attacker abuses Qubole's trusted AWS cross-account role to raid customer secrets

QuboleAWS cross-account rolesWiz

WhatSomeone used Qubole's AWS credentials to assume customers' cross-account IAM roles and pull secrets and S3 buckets.

Details and sources
Wiz and Sentra report that someone obtained Qubole (Idera)'s own AWS account credentials and used them to assume customers' cross-account IAM roles — most of them unused but still trusting Qubole's account — pulling Secrets Manager, SSM and Lambda-environment secrets (including customers' own AWS keys, Entra ID and Okta client secrets) and, in some cases, entire S3 buckets. No attribution has been established, and researchers say the targeting doesn't match known state-actor patterns. The Hacker News (source)
🟠 High impact

Pro-Russian hackers dox 18 European officials including Italy's foreign minister

NoName057(16)doxxingItaly

WhatThe pro-Russian group published personal data on about 18 European officials and journalists, including Italy's foreign minister.

Details and sources
NoName057(16) published personal data on roughly 18 European officials and journalists — including Italian Foreign Minister Antonio Tajani and Eurojust/Europol/BKA leadership — on the dark web. Italian PM Giorgia Meloni blamed pro-Russian hackers; Italy says G7 foreign ministers are now set to discuss countermeasures against the campaign. Al Jazeera (source) · Reuters (source)
‹ 09/11/202611/11/2026 ›