Koffei

Daily CyberNews · 16/10/2026

‹ 15/10/202619/10/2026 ›
Today in 30 seconds
  • 🔴 Patch Cisco Secure Email Gateway by Oct 17: CVE-2026-76519 is exploited and in KEV.
  • 🔴 Update Nextcloud richdocumentscode to 26.4.303: CVE-2026-110447 is used to backdoor servers.
  • 🔴 Ivanti Connect Secure CVE-2026-104935 and Windows CVE-2026-96951 KEV deadlines fall today, Oct 16.
  • 🔴 Rotate CI/CD secrets: GhostAction compromised about 1,900 GitHub accounts and targeted 30,000 repositories.
  • 🟠 Upgrade Next.js to 16.4.1 or 15.5.28 for two critical unauthenticated RCEs including CVE-2026-96889.

Vulnerabilities 7

🔴 Exploited or active

Cisco Secure Email Gateway zero-day under active attack

CVE-2026-76519CVSS 9.8Cisco Secure Email Gatewayin KEV

WhatAn unauthenticated path traversal lets a remote sender write files to AsyncOS when File Reputation/AMP is on, possibly giving root; exploited in the wild.

DoUpdate AsyncOS to 15.5.5-014 or 16.5.0-780 by Oct 17; 16.0 has no fix and must migrate.

Details and sources
An unauthenticated path-traversal flaw, CVE-2026-76519 (CVSS 9.8, CWE-23), lets a remote sender write arbitrary files to AsyncOS for Secure Email Gateway when File Reputation/AMP is enabled, potentially leading to root command execution. Cisco confirmed in-the-wild exploitation; CISA added it to the KEV catalog on Oct 14 with a federal patch deadline of Oct 17. Fixed in AsyncOS 15.5.5-014 and 16.5.0-780 (no 16.0 fix — migration required). The Hacker News (source) · Canadian Centre for Cyber Security (source)
🔴 Exploited or active

Ivanti Connect Secure zero-day exploited, CISA emergency order

CVE-2026-104935Ivanti Connect Securein KEV

WhatThe Ivanti Connect Secure auth bypass was exploited as a zero-day, and Volexity found three exploit chains with one still open as of Oct 14.

DoUpdate to 25.1.1.2, 25.1.3.0/25.1.3.1 or 22.8R2.5; the federal deadline is Oct 16.

Details and sources
CVE-2026-104935 (CVSS 8.6, CWE-290, auth bypass by spoofing) in Ivanti Connect Secure VPN appliances was exploited as a zero-day against a limited set of customers before a patch existed. Volexity found three distinct exploit chains, one still open as of Oct 14; activity is assessed as likely criminal rather than espionage. Fixed in 25.1.1.2, 25.1.3.0/25.1.3.1, 22.8R2.5. CISA added it to KEV on Oct 13 and ordered federal agencies to patch by Oct 16 — the sixth Ivanti bug added to KEV in 2026. Rapid7 (source) · The Record (source)
🔴 Exploited or active

Microsoft October Patch Tuesday — record CVE count, one actively exploited zero-day

CVE-2026-96951Patch Tuesdayin KEV

WhatOctober Patch Tuesday fixed 1,400+ flaws, including the exploited Windows Shell MotW bypass and critical AD and Outlook RCEs triggered by viewing email.

DoInstall October Patch Tuesday updates by the Oct 16 KEV deadline, also covering CVE-2026-108398 and CVE-2026-92707.

Details and sources
Microsoft's Oct 13 release fixed over 1,400 vulnerabilities (69 critical), including three zero-days. CVE-2026-96951 (Windows Shell, CVSS 8.8, Mark-of-the-Web bypass) is confirmed actively exploited and was added to CISA KEV on Oct 13 with an Oct 16 federal deadline; CVE-2026-108398 (.NET Framework, WDAC bypass) and CVE-2026-92707 (MessagePack-CSharp DoS) were publicly disclosed pre-patch but not confirmed exploited. Also notable: a critical unauthenticated Active Directory RCE and five Outlook RCEs rated 9.8, each triggerable just by viewing a malicious email. CrowdStrike (source) · The Record (source)
🟠 High impact

Next.js ships emergency fix for two critical unauthenticated RCEs

CVE-2026-96889CVSS 9.5Next.jsVercel

WhatNext.js fixed an unauthenticated RCE in use cache/Cache Components and an Image Optimization RCE via a libvips/sharp SVG flaw; no exploitation reported.

DoUpgrade Next.js to 16.4.1 or 15.5.28; for the image flaw, the workaround is sharp 0.35.5 or later.

Details and sources
Vercel released Next.js 16.4.1 and 15.5.28 out-of-band on Oct 14 to fix a critical unauthenticated RCE in use cache/Cache Components (no workaround) and CVE-2026-96889 (CVSS 9.5, RCE in the Image Optimization API via an upstream libvips/sharp flaw when processing SVGs on glibc-based Linux; workaround is sharp ≥0.35.5), plus a high-severity Server Actions DoS. No in-the-wild exploitation reported at release. Next.js (source)
🟠 High impact

Drupal patches no-auth-required Field UI RCE path

CVE-2026-108097Drupal coreField UI

WhatA Drupal core Field UI flaw has at least one unauthenticated exploitation path through contributed or custom modules; no active exploitation confirmed.

DoUpdate Drupal to 10.6.19, 11.3.19, 11.4.10 or 12.0.0-beta2, or disable the field_ui module meanwhile.

Details and sources
CVE-2026-108097 in Drupal core's Field UI module is not directly exploitable in core alone, but Drupal confirms at least one exploitation path via contributed/custom modules requires no authentication. No CVSS score has been published yet. Affects Drupal ≥8.9.0 and multiple 10.x/11.x/12.0-alpha branches; fixed in 10.6.19, 11.3.19, 11.4.10, and 12.0.0-beta2. Interim mitigation is to disable the field_ui module; no active exploitation confirmed. Vulert (source)
🔴 Exploited or active

Nextcloud/Collabora RCE used to backdoor servers

CVE-2026-110447CVSS 9.2NextcloudCollabora CODE

WhatA command injection in Nextcloud's built-in Collabora CODE app lets unauthenticated attackers drop a PHP web shell via proxy.php; three admins confirmed compromise.

DoUpdate richdocumentscode to 26.4.303 and check for PHP web shells; standalone Collabora servers are unaffected.

Details and sources
A command-injection flaw, CVE-2026-110447 (provisional CVSS 9.2, CWE-78), in Nextcloud's "Collabora Online – Built-in CODE Server" app lets an unauthenticated attacker drop a PHP web shell via proxy.php while the local document server is down (e.g., during restart). At least three admins have confirmed compromise, with attackers searching for password-manager files and crypto wallets. Standalone Collabora servers are unaffected; fixed in richdocumentscode 26.4.303. The Hacker News (source)
🟡 Worth knowing

NGINX and SAP patch days

CVE-2026-15176CVE-2026-97205NGINXSAP

WhatNGINX fixed five flaws including two dating back to 2009, and SAP's patch day fixed 20 issues; no known exploitation.

DoUpdate NGINX to 1.31.7 or 1.30.6.

Details and sources
NGINX 1.31.7/1.30.6 (Oct 14) fixed five flaws including two bugs dating back to 2009 (CVE-2026-15176, CVE-2026-97205); no known in-the-wild exploitation. SAP's Oct 13 Security Patch Day fixed 20 issues, headlined by a maximum-severity (CVSS 10.0) SSRF in the SAP Cloud SDK for Python that could let an attacker steal OAuth2 client credentials on multi-tenant BTP apps; SAP reports no active exploitation or public PoC. Daily Security Review (source)

Breaches / Incidents 8

🟠 High impact

BAE Systems cyberattack disrupts US federal bankruptcy court notices

BAE SystemsUS courtsBankruptcy Noticing Center

WhatA cyberattack on BAE Systems took down bnc.uscourts.gov and halted nationwide bankruptcy notices; data theft is not confirmed.

Details and sources
A cyberattack on BAE Systems, which runs the Bankruptcy Noticing Center for the US federal judiciary, took down bnc.uscourts.gov and related services, halting nationwide bankruptcy-notice production and distribution. The Administrative Office of the US Courts confirmed the outage; BAE is investigating, with no confirmation yet on data theft. BleepingComputer (source)
🟠 High impact

Meta discloses GitHub breach affecting ~1,500 employees

MetaGitHubstolen credentials

WhatAttackers used a stolen employee's credentials to take source code and business data from Meta's internal GitHub, affecting about 1,500 employees.

Details and sources
Attackers used a stolen Meta employee's credentials to access internal GitHub repositories, taking mostly source code and business data plus a small amount of personal data (largely synthetic/test data). Meta is notifying about 1,500 mostly US-based employees, found no evidence of product compromise, and has rotated secrets and contacted law enforcement. The Arabian Post (source)
🟠 High impact

UK Legal Aid Agency breach confirmed far larger than first disclosed

UKLegal Aid Agencybreach scope

WhatForensics show the Legal Aid Agency breach includes records back to 1978, with attackers claiming data on over 2 million people.

Details and sources
New forensic analysis shows roughly 250,000 stolen records predate 2007, with the earliest verified record from 1978 — well beyond the original "since 2010" estimate given to Parliament. Stolen data spans names, dates of birth, National Insurance numbers, case and financial-assessment records, partial card data, and probation records; attackers claim to hold data on over 2 million people, though none has been confirmed published. The UK government maintains its no-ransom policy and is working with the NCA, NCSC, and ICO. The Record (source)
🟡 Worth knowing

Burnaby School District (Canada) confirms ransomware data theft

CanadaBurnaby School Districtransomware

WhatThe Burnaby school board confirmed a September ransomware attack stole personal data of students, parents and staff.

Details and sources
The British Columbia school board confirmed that attackers stole personal information belonging to students, parents, and staff in a ransomware attack that occurred in September 2026; full scope has not yet been detailed. CTV News (source)
🟠 High impact

"n0n" extortion group claims mass breach of 38 shipping and logistics firms

n0nshippinglogisticsCyberLogitec

WhatThe n0n group claims data from 38 shipping and logistics firms in 18 countries, possibly via one shared app tied to CyberLogitec; unverified.

Details and sources
A data-theft extortion group listed 38 shipping, port, and logistics companies across 18 countries — including Ocean Network Express, CULines, and Santos Brasil — on its leak site, demanding $2.4M–$14M in Bitcoin and claiming EDI archives, container manifests, and financial records. The claims are unverified; analysts suggest the small, app-specific datasets per victim could point to one shared vulnerable application rather than 38 separate breaches, possibly tied to CyberLogitec software. Splash247 (source)
🟡 Worth knowing

Peruvian insurer Interseguro confirms breach, "Direwolf" extortion claim

PeruInterseguroDirewolf

WhatPeruvian insurer Interseguro contained unauthorized access detected Oct 7, while Direwolf claims about 2.5TB of stolen data.

Details and sources
Interseguro detected unauthorized system access on Oct 7, contained it, and notified affected customers. Extortion group Direwolf claims roughly 2.5TB of stolen data including internal documents and source code; the insurer says exposed data is mainly IDs, contact details, and policy numbers, with policies and payments operating normally. Peru's SBS regulator is reviewing compliance. Rio Times (source)
🟡 Worth knowing

Watchlist update — Albert Heijn/Etos/Gall & Gall

NetherlandsAlbert HeijnAhold Delhaize

WhatAhold Delhaize is restocking and restarting systems while online ordering stays down; attack traffic was reportedly traced partly to US, Polish, Moldovan and Turkish servers.

Details and sources
Ahold Delhaize has begun restocking Dutch stores and restarting IT systems, with several distribution centres delivering again and digital shelf labels restored; online ordering, delivery of existing orders, and customer service remain down. The company still has not explicitly confirmed a cyberattack, calling it a "cybersecurity incident," and says it has found no sign customer data was taken. Dutch security sources told media that suspicious attack traffic was identified and traced partly to servers in the US, Poland, Moldova, and Turkey, possibly including compromised Ahold Delhaize US infrastructure — but no attacker has been named. DutchNews (source) · NH Nieuws (source)
🟠 High impact

Watchlist update — French supermarket refrigeration attack

FranceDanfossCopeland700 supermarkets

WhatThe refrigeration attack hit about 700 French supermarkets through end-of-life Danfoss AK-SM 800A and Copeland XWeb Pro remote-control modules.

Details and sources
France's Finance Minister told Parliament the attack hit roughly 700 of the country's 10,000 supermarkets (Groupement Mousquetaires/Intermarché ~300 stores, Coopérative U ~80 of 2,000 stores); the government says there is no risk to food supply. ANSSI said it is "aware of numerous compromises" involving refrigeration remote-control modules made by Danfoss (AK-SM 800A) and Copeland (XWeb Pro) — both vendors say the affected units are older, end-of-life product generations no longer receiving security updates. No attacker has been named; the Paris prosecutor's cybercrime unit is investigating alongside European counterparts. News4Jax (source)

Malware / Ransomware 5

🔴 Exploited or active

GhostAction GitHub supply-chain campaign hits ~1,900 accounts

GhostActionGitHub Actions1,900 accounts

WhatGhostAction pushed malicious workflows such as security-audit.yml from about 1,900 hijacked accounts to steal CI/CD, cloud and API secrets.

DoRotate CI/CD secrets and look for workflows like security-audit.yml; unrotated secrets caused repeat victims.

Details and sources
A credential-theft campaign dubbed GhostAction used stolen personal access tokens and infostealer-log credentials to push malicious GitHub Actions workflow files (e.g. "security-audit.yml") designed to exfiltrate CI/CD secrets, cloud credentials, and npm/PyPI/DockerHub/AI API keys. This wave compromised roughly 1,900 accounts and attempted to pull over 3,100 secrets from about 30,000 repositories; GitHub began blocking the malicious workflow on Oct 9, limiting later waves. Some victims were repeat targets from earlier 2025/2026 waves due to unrotated secrets. The Hacker News (source) · BleepingComputer (source)
🟠 High impact

Hirudo — new Android banking trojan targets Indonesian banks

HirudoAndroid banking trojanIndonesia

WhatHirudo uses Accessibility to replace genuine bank apps with patched copies, fakes integrity checks and steals credentials, PINs and SMS.

Details and sources
ThreatFabric identified Hirudo, an Android banking trojan distributed via malicious loader apps disguised as shopping, adult-content, and short-drama apps. It abuses Accessibility permissions to uninstall a genuine bank app and install a patched copy via a hooking engine, fakes signing-certificate and Play Integrity checks, enables live screen streaming, and harvests banking credentials, PINs, and SMS. It impersonates at least seven Indonesian banks, with secondary targeting in China, India, Mexico, Morocco, and Thailand. BleepingComputer (source)
🔴 Exploited or active

Fake Arch Linux AUR packages impersonating yay/paru deliver root-privilege malware

Arch LinuxAURyayparusupply chain

WhatMalicious AUR packages impersonating the yay and paru helpers deliver malware with root privileges.

DoDo not install AUR packages impersonating yay or paru, and check systems that did.

Details and sources
Malicious Arch User Repository packages impersonating the popular "yay" and "paru" AUR helper tools were posted to deliver malware with root privileges, continuing a pattern of AUR supply-chain abuse (an earlier wave compromised 400+ packages to push a Linux rootkit and infostealer). Full technical detail on this specific wave is still limited. BleepingComputer (source)
🟡 Worth knowing

ThreatsDay roundup — new malware families and a DNS-based wallet drainer

EtherHidingNoirAokiXWorm

WhatA roundup flagged a WordPress backdoor using EtherHiding, the DNS-based Noir wallet drainer, and stealers like Aoki targeting AI API keys.

Details and sources
The Hacker News' Oct 15 bulletin flagged a self-healing WordPress backdoor that hides its C2 address in an Ethereum smart contract (EtherHiding) to resist removal; Noir, a DNS-powered crypto-wallet drainer that reads DNS TXT records via DNS-over-HTTPS and needs no seed phrase; and several new or updated malware families including CikaBika Bot (Rust RAT with Discord C2), KGB RAT, BotHelper RAT, Warden Stealer (targets AI coding-tool configs), CastleStealer, Aoki (targets Claude/Codex/Cursor/Gemini API keys and session tokens), and an expanded 46-plugin surveillance version of XWorm. The Hacker News (source)
🔴 Exploited or active

Fake AI-tool download pages spread infostealers

Fake AI tool sitesinfostealersmshta

WhatSponsored results and fake download pages for Claude, ChatGPT, Copilot and Gemini trick users into pasting commands that run mshta, HTA and PowerShell stealers.

DoHunt for mshta.exe launching HTA files followed by encoded PowerShell on user endpoints.

Details and sources
Campaigns are using sponsored search results, legitimate hosting, and counterfeit download sites impersonating Claude, ChatGPT, GitHub Copilot, and Gemini to trick users into running clipboard-copied commands that chain mshta.exe into a malicious HTA and encoded PowerShell, ultimately stealing browser sessions and credentials; US organizations account for 16% of country-attributed targets. The Hacker News (source)

Threat Actor Activity 7

🟡 Worth knowing

DOJ/FBI dismantle North Korean "laptop farm" network in Texas

North KoreaIT workerslaptop farmsTexas

WhatTwo US nationals were arrested and ten China-based operators charged for running laptop farms placing DPRK IT workers at 70+ US companies.

Details and sources
Two US nationals were arrested and charged with wire fraud, money laundering, and identity theft for helping DPRK IT workers steal American identities to land remote jobs at 70+ US companies, funneling proceeds to North Korea. FBI raids on 40+ suspected "laptop farm" residences seized about 187 devices; 25 Americans' identities were compromised and at least $3M in revenue was generated. Ten China-based operators were also charged. BleepingComputer (source) · DOJ (source)
🟡 Worth knowing

North Macedonia arrests alleged administrator of OLVX cybercrime marketplace

OLVXNorth Macedoniaarrest

WhatNorth Macedonia police and the FBI arrested the alleged administrator of the OLVX marketplace selling stolen data, credentials and RDP access.

Details and sources
Police arrested a suspect believed to administer OLVX, a marketplace active 2023–2025 with over 1,500 seller accounts trading stolen data, credentials, phishing kits, and RDP access, in a joint operation with the FBI. The same suspect allegedly ran a separate Telegram-based refund-fraud toolkit group. The Record (source)
🟡 Worth knowing

Microsoft disrupts Chinese-language phishing-as-a-service ecosystem

MicrosoftYAM StoreCoGUIIron Man

WhatMicrosoft took down three Chinese-language phishing platforms that sent over 750 million malicious messages and disabled 20,000+ YAM Store accounts.

Details and sources
Microsoft took down three linked platforms — YAM Store (an initial-access broker selling compromised Microsoft 365/Google accounts), CoGUI (brand-impersonation phishing kits heavily used against Japan), and Iron Man (phishing-kit-as-a-service with over 2,300 lure pages). Since January the platforms reportedly sent over 750 million malicious messages and impersonated 75,000+ organizations across 150+ countries; Microsoft obtained a US court order, working with Japan's National Police Agency, to seize domains and disabled over 20,000 YAM Store accounts. The Record (source)
🔴 Exploited or active

Silent Lynx (Kazakhstan-linked) exploits WinRAR flaw against CIS governments

Silent LynxCVE-2025-8088WinRARCIS governments

WhatSilent Lynx exploited WinRAR CVE-2025-8088 in April 2026 to plant the PickleShell backdoor against government ministries in nine CIS countries.

DoHunt Windows Startup folders for the PickleShell backdoor on government systems in Central Asia.

Details and sources
Cisco Talos (tracking it as UAT-6275) and Recorded Future (TAG-193) detailed Silent Lynx, active since 2020 with a suspected link to YoroTrooper. In April 2026 the group used WinRAR vulnerability CVE-2025-8088 to drop a custom backdoor, PickleShell, into the Windows Startup folder — its first known use of a software exploit rather than pure social engineering. Targets spanned government ministries in nine CIS/Central Asian countries, with medium-confidence Kazakhstan attribution based on Russian-language artifacts and malware testing timed to Kazakhstan working hours. The Hacker News (source) · The Record (source)
🟠 High impact

Citizen Lab: Candiru spyware reached Al Jazeera devices via zero-click malicious ads

CandiruAl Jazeerazero-click ads

WhatCitizen Lab tied zero-click spyware delivered via targeted malicious ads to at least 45 Al Jazeera devices to Candiru with high confidence.

Details and sources
Citizen Lab, with ad-security firm Confiant, attributed with high confidence a zero-click spyware campaign against at least 45 Al Jazeera devices (2024–2025) to Israeli spyware maker Candiru. Malicious ads served through an ad-tech intermediary silently fingerprinted devices and served exploit code only to intended targets, with no click required — described as the first public technical confirmation of mercenary spyware delivered via a targeted ad network. Several ad platforms have since cut off the implicated ad-tech firm's access. The Hacker News (source)
🟡 Worth knowing

Brooklyn DA disrupts crypto-investment scam network targeting Russian-speaking New Yorkers

Brooklyn DAFBIcrypto-investment scam

WhatThe Brooklyn DA and FBI seized 241 domains and 112 servers of a crypto scam targeting Russian-speaking New Yorkers.

Details and sources
The Brooklyn DA's office and FBI seized 241 domains and 112 servers tied to a crypto-investment fraud ring that targeted Russian-speaking Brooklyn residents via Russian-language social media ads and fake "investment adviser" and "recovery" scams. Infrastructure traced to Ukraine and other Eastern European countries; victims lost millions, with roughly $300,000 recovered so far. The Record (source)
⚪ Context

Watchlist update — Ukrainian drone strikes on Yandex data centers

Yandexdrone strikesbot traffic

WhatResearchers say the 94% drop in Yandex automated traffic likely reflects its own crawlers going offline, not reduced troll-farm activity.

Details and sources
No new strikes have been reported since the third hit on Oct 11 in the Vladimir region, after which Yandex fully suspended that site. President Zelenskyy's claim that Russian bot/troll activity dropped as a result remains disputed: Cloudflare data did show automated traffic from Yandex's network fall roughly 94% around Oct 11, but researchers at Georgetown, Antibot4Navalny, and other groups say most of the drop likely reflects Yandex's own web crawlers going offline rather than troll-farm activity, and found no corresponding drop in tracked Kremlin-aligned or AfD-related accounts. Yellow.com (source)
‹ 15/10/202619/10/2026 ›