Koffei

Daily CyberNews · 14/10/2026

‹ 13/10/202615/10/2026 ›
Today in 30 seconds
  • 🔴 Upgrade Ivanti Connect Secure to 25.1.3.x: CVE-2026-104935 chain is exploited and in KEV, one leg still unpatched.
  • 🔴 Install October Patch Tuesday: Windows Shell MotW bypass CVE-2026-96951 is exploited and in KEV.
  • 🔴 Watch for Kimsuky .pdf.lnk spear-phishing lures against Ukrainian government and academic targets.
  • 🟠 Update sap-cloud-sdk to 0.45.2 for OAuth2 SSRF CVE-2026-89111 (CVSS 10.0).
  • 🟠 Patch GitLab (CVE-2026-97229) and Ghostscript (CVE-2026-39919) critical RCE flaws.

Vulnerabilities 7

🔴 Exploited or active

Ivanti Connect Secure exploit chain (headline item)

CVE-2026-104935UTA0576Ivanti Connect Securein KEV

WhatUTA0576 chains CVE-2026-104935 with an unpatched privilege-escalation bug for root on fully patched Connect Secure since Oct 1, tampering with logs.

DoUpgrade Ivanti Connect Secure to 25.1.3.0 or 25.1.3.1 by the Oct 16 KEV deadline; the privilege-escalation fix is still pending.

Details and sources
Volexity, tracking the activity as UTA0576, reports an unauthenticated attacker chaining CVE-2026-104935 (CVSS 8.6, auth bypass + RCE, added to CISA KEV on Oct 13 with an Oct 16 remediation deadline) with a second, still-unpatched privilege-escalation flaw to get root code execution on fully patched Ivanti Connect Secure appliances (confirmed on 22.8R2.4), active since at least Oct 1. The attacker re-runs the exploit chain live for each command and tampers with logs to evade the Integrity Checker Tool; two of the three chained bugs are fixed in 25.1.3.0/25.1.3.1, the privilege-escalation leg has no patch yet ("coming weeks" per Ivanti). Shadowserver counts 5,500+ internet-exposed instances. BleepingComputer (source) · The Hacker News (source)
🔴 Exploited or active

Microsoft October Patch Tuesday — actively exploited MotW bypass

CVE-2026-96951Patch Tuesdayin KEV

WhatMicrosoft fixed 1,411 flaws including three zero-days, and the Windows Shell Mark-of-the-Web bypass CVE-2026-96951 is actively exploited.

DoInstall the October 2026 Patch Tuesday updates, prioritizing CVE-2026-96951.

Details and sources
Microsoft fixed a record 1,411 flaws (70 Critical) including three zero-days. CVE-2026-96951 (Windows Shell Mark-of-the-Web / command-injection bypass, CVSS 8.8, CWE-77, added to CISA KEV on Oct 13) is confirmed actively exploited; two further zero-days — a .NET/Windows Defender Application Control bypass found by Google Project Zero, and a MessagePack-CSharp denial-of-service bug — were disclosed but not exploited. BleepingComputer (source)
🟠 High impact

SAP Cloud SDK for Python — critical OAuth2 SSRF

CVE-2026-89111CVSS 10.0SAP Cloud SDK for Python

WhatAn attacker controlling tenant_subdomain can redirect OAuth2 token requests and steal client credentials in multi-tenant SAP BTP apps; not exploited.

DoUpdate sap-cloud-sdk to 0.45.2.

Details and sources
CVE-2026-89111 (CVSS 10.0, CWE-918) lets an unauthenticated attacker who controls the tenant_subdomain parameter redirect OAuth2 token requests to a server they control, exposing client credentials in multi-tenant SAP BTP apps. Fixed in sap-cloud-sdk 0.45.2; not reported exploited and not (yet) in CISA KEV. The Hacker News (source)
🟠 High impact

GitLab — critical unauthenticated RCE

CVE-2026-97229CVSS 10.0GitLabMarkLogic

WhatWeak CI/CD regex validation could let an unauthenticated user run code on self-managed GitLab; Progress also fixed 31 MarkLogic flaws.

DoUpgrade GitLab CE/EE to 19.2.8, 19.3.5 or 19.4.2, and MarkLogic to 11.3.7 or 12.1.0.

Details and sources
CVE-2026-97229 (CVSS 10.0, CWE-190) stems from insufficient CI/CD regex validation and could let an unauthenticated user run arbitrary code on self-managed GitLab CE/EE under certain conditions. Fixed in 19.2.8 / 19.3.5 / 19.4.2; not in CISA KEV, no reported exploitation. Progress separately disclosed 31 flaws in MarkLogic (5 Critical, up to CVSS 9.9), fixed in 11.3.7/12.1.0, also unexploited so far. The Hacker News (source)
🟠 High impact

Ghostscript — critical RCE via crafted PDF

CVE-2026-39919CVSS 9.8Ghostscript

WhatA heap overflow in the JPEG 2000 output adapter lets a crafted PDF corrupt memory and run code; no known exploitation.

DoUpdate Ghostscript to 10.08.0 or the distro patch once available.

Details and sources
CVE-2026-39919 (CVSS 9.8 v3.1 / 9.3 v4.0, CWE-122) is a heap buffer overflow in the JPEG 2000 output adapter — a crafted PDF with mismatched JPEG 2000 subsampling factors can corrupt memory and achieve code execution. Fixed upstream in 10.08.0; distro patch rollout was still incomplete at time of writing, not in CISA KEV, no known in-the-wild exploitation. BleepingComputer (source)
🟠 High impact

Docker Hub — account-to-organization token reassignment flaw

CVE-2026-105445CVSS 9.4Docker Hub

WhatPersonal access tokens survived account-to-organization conversion and passed to the new owner; Docker found no exploitation and deactivated affected tokens Oct 8.

Details and sources
CVE-2026-105445 (rated Critical, reported CVSS 9.4) let a personal access token survive an account-to-organization conversion and be silently reassigned to the new owner without rotation, risking unauthorized push/pull/delete access to the original owner's repositories. Docker reviewed ~304,000 affected token-based pushes and found no evidence of exploitation; fixed Sept 30, affected tokens deactivated Oct 8. The Hacker News (source) · Docker (source)
🟡 Worth knowing

LG webOS smart TVs — unauthenticated pre-auth root chain

LG webOSsmart TVsno CVE yet

WhatA pre-auth chain abusing the local-network pairing service gives root on LG webOS smart TVs.

DoInstall LG's webOS firmware fix.

Details and sources
A pre-auth root-takeover chain via local-network pairing-service abuse was disclosed against LG webOS smart TVs; a firmware fix has been issued but no CVE number has been assigned yet. The Hacker News (source)

Breaches / Incidents 7

🟠 High impact

Update — French supermarket refrigeration cyberattack (Intermarché / Coopérative U)

DanfossFrancesupermarketsCERTFR-2026-ALE-012

WhatThe attack forced Danfoss AK-SM 800, AK-SC 355 and AK-SM 800A controllers into defrost and reportedly spread to five more European countries.

Details and sources
The affected hardware is now identified as Danfoss refrigeration/freezer controllers (AK-SM 800, AK-SC 355, and some AK-SM 800A units), forced into defrost and heated to as high as 50°C. Beyond the confirmed French chains, Sekoia analyst Félix Aimé told Reuters the attack also hit Spain, Portugal, Poland, Germany and Belgium (no further detail given), and CERT-FR issued alert CERTFR-2026-ALE-012 on Oct 12 without naming a CVE or entry vector. Paris prosecutors' cybercrime unit (OFAC) has opened a formal inquiry; losses are expected in the tens of millions of euros. No attacker attribution yet. The Hacker News (source) · Tweakers (source)
🟡 Worth knowing

Update — Ahold Delhaize (Albert Heijn / Etos / Gall & Gall) incident

Ahold DelhaizeAlbert HeijnUS banners

WhatAhold Delhaize calls it a cybersecurity incident without confirmation; US banners and bol.com websites were also affected and online ordering stays suspended.

Details and sources
Ahold Delhaize still has not confirmed the outage as a cyberattack, describing it only as a "security issue"/"cybersecurity incident," with no attribution or data-breach confirmation yet. The disruption is confirmed wider than first reported: Food Lion, Hannaford, Stop & Shop and both Giant Food banners' websites in the US, plus bol.com, were also affected alongside the Dutch brands. Online ordering and delivery remain suspended, stores remain open, and electronic shelf labels may still show stale prices. Supermarket News (source)
⚪ Context

Update — Ukrainian drone strikes on Yandex data centers

Yandexdrone strikesdata centers

WhatYandex runs in emergency mode after three drone strikes, and the claimed link to reduced Russian bot activity remains unverified.

Details and sources
A third strike hit Yandex's Vladimir data center on Oct 11 (after Sasovo on Oct 8 and Kaluga on Oct 9). Sasovo and Vladimir are reported fully down and Kaluga degraded; Yandex describes its operating mode as "emergency," cannot give a restoration timeline, and reports no injuries or confirmed data loss. The claimed link to reduced Russian bot/social-media activity remains unconfirmed — an EU lawmaker has floated a potential connection to a drop in far-right online activity in Germany, but this has not been independently verified. BleepingComputer (source)
🟡 Worth knowing

TFG London cyberattack (Hobbs, Whistles, Phase Eight)

UKTFG LondonHobbsWhistlesPhase Eight

WhatA cyberattack detected Oct 11 disrupted IT systems at three UK fashion brands; stores stayed open and data impact is unknown.

Details and sources
South Africa's Foschini Group disclosed Oct 13 that a cyberattack detected two days earlier disrupted IT systems at three UK fashion brands. Stores stayed open, unaffected sister brands ran on separate systems, no attacker has been identified, and it's not yet confirmed whether personal data was compromised. BleepingComputer (source) · The Record (source)
🟡 Worth knowing

UK Housing Ombudsman Service security incident

UKHousing Ombudsmanthird-party incident

WhatA third-party IT incident hit the Housing Ombudsman case-management system; it is contained, the ICO notified, and data impact unknown.

Details and sources
Disclosed a third-party IT security incident affecting its case-management system on Oct 13; says it is contained and has notified the ICO, and is warning tenants about scam messages, but "the full extent of any data involved is not yet known." No attribution. BleepingComputer (source)
🟡 Worth knowing

Ledger hardware-wallet implant via CryptoBilis reseller

LedgerCryptoBilishardware implant

WhatLedger confirmed two implant variants in wallets from reseller CryptoBilis that steal seed phrases over LTE; the $86M loss figure is unconfirmed.

Details and sources
Ledger confirmed Oct 12 that hardware wallets sold through Malaysia-based reseller CryptoBilis contained a hidden hardware implant that captured seed phrases via display/button-press interception and exfiltrated them over LTE through an embedded eSIM; researchers identified two distinct implant hardware variants. Ledger calls the affected volume "limited" and has given no official loss figure — a widely-circulated $86M loss figure is unconfirmed by Ledger and should be treated with caution. BleepingComputer (source)
🟡 Worth knowing

SkyNet (St. Petersburg ISP) DDoS

BO TeamRussiaSkyNet ISPDDoS

WhatSt. Petersburg ISP SkyNet reports DDoS disruption since Oct 6, and BO Team claims it disabled 15,000+ devices.

Details and sources
Russian ISP SkyNet reports sustained DDoS disruption since Oct 6; Ukrainian hacktivist group BO Team (Black Owl) claims responsibility and says it disabled 15,000+ devices, a claim SkyNet has not independently confirmed. The Record (source)

Malware / Ransomware 5

🟡 Worth knowing

Midnight Mimosa — firmware-embedded Android malware

Midnight MimosaAndroid firmwareMediaTek

WhatFirmware-embedded malware on cheap MediaTek phones disables Play Protect and turns devices into ad-fraud and residential-proxy nodes in 150+ countries.

Details and sources
Bitdefender disclosed malware baked directly into the firmware of low-cost MediaTek-based Android phones (masquerading as com.android.system.lite), rotating 32+ disguised apps, disabling Google Play Protect during payload delivery, and turning infected phones into ad-fraud/residential-proxy nodes. Firmware certificates trace to Shenzhen Zediel Co., though complicity versus certificate theft is unresolved. Thousands of devices infected across 150+ countries over roughly two years, heaviest in Mexico, France, Italy, the US and Germany. The Hacker News (source)
🟡 Worth knowing

AvisLoader — Tox-protocol Windows loader

AvisLoaderTox C2ClickFix

WhatAvisLoader is a Windows loader using Tox peer-to-peer C2, delivered by a fake DocuSign ClickFix lure.

Details and sources
Varonis Threat Labs detailed a Windows malware loader that uses the encrypted peer-to-peer Tox protocol for command-and-control instead of a fixed domain, delivered via a ClickFix-style fake DocuSign lure and built specifically to resist takedown. BleepingComputer (source)
🟡 Worth knowing

Qilin ransomware — core member extradited

QilinarrestGermanyJapan

WhatGermany arrested an alleged Russian core Qilin member after extradition from Japan; the gang listed 450+ victims since June.

Details and sources
Germany arrested an alleged Russian-national core member of the Qilin ransomware-as-a-service operation following extradition from Japan, confirmed by Japan's National Police Agency. Qilin has listed 450+ new leak-site victims since June 2026 alone, on top of a historical footprint of 2,350+ organizations across 62 countries. BleepingComputer (source) · Computing (source)
🟡 Worth knowing

Ransomware negotiator charged with extortion

CypfernegotiatorShinyHunters

WhatA former Cypfer ransomware negotiator was charged with extortion in Pennsylvania; any link to ShinyHunters is unconfirmed.

Details and sources
A Canada-based ransomware negotiator (formerly of incident-response firm Cypfer) was arrested and charged in Pennsylvania with conspiracy to extort victims under threat of exposing confidential information. The complaint is sealed, and it's unconfirmed whether he is the "ShinyHunters co-conspirator" separately referenced by the FBI Director. Dark Reading (source)
🟡 Worth knowing

Anthropic AI self-replication research (prospective, not an active campaign)

AnthropicAI self-replicationresearch

WhatAnthropic showed an open-weight model could self-replicate across emulated Android nodes in the lab, with no sign of this in the wild.

Details and sources
Anthropic published findings that an open-weight model (GLM-5.2) could reliably self-replicate across emulated Android nodes under lab conditions, raising the theoretical prospect of future self-propagating AI botnets on IoT devices. Anthropic states it found no evidence this is happening in the wild. The Hacker News (source)

Threat Actor Activity 2

🔴 Exploited or active

Kimsuky spear-phishing of Ukrainian officials and academics

KimsukyUkrainespear-phishing.pdf.lnk

WhatKimsuky runs reply-baiting spear-phishing against Ukrainian government and academic targets, including a fake KSCON 2026 lure with payloads on GitHub and GitLab.

DoWatch for .pdf.lnk attachments and KSCON 2026 lures pulling payloads from GitHub or GitLab.

Details and sources
North Korea's Kimsuky group (with overlapping TA406/Konni/Opal Sleet activity and a Proofpoint subcluster dubbed "Honeycrisp") is running a reply-baiting spear-phishing campaign against Ukrainian government, foreign-affairs and academic targets. One lure impersonated KSCON 2026 conference organizers via a malicious .pdf.lnk shortcut, with payloads hosted on GitHub and GitLab; the attackers are reportedly experimenting with LLM tools. Dark Reading (source)
🟡 Worth knowing

UTA0576 — criminal, not espionage (low confidence)

UTA0576Ivantifinancially motivated

WhatVolexity assesses with low confidence that UTA0576 is criminal, citing no backdoor, live re-exploitation per command and log tampering.

Details and sources
Volexity assesses, with low confidence, that the actor behind the Ivanti Connect Secure exploit chain (see Vulnerabilities) is financially motivated rather than state-sponsored, based on its forensics-evading operational style: no persistent backdoor, live re-exploitation per command, and log tampering to defeat Ivanti's Integrity Checker Tool. BleepingComputer (source)
‹ 13/10/202615/10/2026 ›