Koffei

Daily CyberNews · 29/10/2026

‹ 28/10/202630/10/2026 ›
Today in 30 seconds
  • 🔴 Patch Cisco ASA/FTD: CVE-2026-76418 is exploited to reload SSL VPN devices.
  • 🔴 Patch PHP and Linux kernels: PHP2Shell plus Fragnesia (CVE-2026-46300, KEV) rooted firewalls and PAN-OS.
  • 🔴 Upgrade Cisco Catalyst SD-WAN: CVE-2026-20127 exploited to plant KeyDrift; no workaround.
  • 🔴 Update Node.js, PHP and .NET for Brotli CVE-2026-111497; patch Adobe Commerce PolyShell, both in KEV.
  • 🔴 Patch VMware ESX: Warp Panda exploited CVE-2026-41709, now in KEV.

Vulnerabilities 7

🟠 High impact

CVE-2026-101767 — Oracle E-Business Suite pre-auth SQLi→RCE

CVE-2026-101767Oracle EBSCVSS 9.8

WhatUnauthenticated SQL injection in the EBS TCF servlet affects versions 12.2.3 to 12.2.15; honeypots saw exploratory probing on Oct 27 but no confirmed compromise.

DoApply Oracle's October 2026 Critical Patch Update to E-Business Suite.

Details and sources
An unauthenticated SQL injection in the TCF servlet (/OA_HTML/AppsTCFServer) of Oracle Application Object Library, CVE-2026-101767 (CVSS 9.8, CWE-287 improper authentication), affects Oracle E-Business Suite 12.2.3–12.2.15. Patched in Oracle's October 2026 Critical Patch Update after Horizon3.ai's disclosure; Defused Cyber recorded exploratory probing against honeypots on Oct 27 (EPSS still low at 0.61%, no confirmed compromise yet) — but EBS has already seen two real mass-exploitation waves this year (CVE-2025-61882/Cl0p and CVE-2026-46817). The Hacker News (source)
🔴 Exploited or active

Brotli decoder flaw and Adobe Commerce "PolyShell" added to CISA KEV

CVE-2026-111497CVE-2026-48356in KEV

WhatCISA added an exploited Brotli decoder out-of-bounds write and the Adobe Commerce "PolyShell" upload flaw, which has breached hundreds of stores, to KEV.

DoInstall Node.js 22.23.4/24.22.1/26.12.1, PHP 8.5.12-1/8.4.27-1/8.3.36-1/8.2.35-1 and .NET 8.0.33/9.0.22/10.0.14; federal deadline Nov 10.

Details and sources
CISA added two actively-exploited bugs to its KEV catalog on Oct 27 with a Nov 10 federal patch deadline: a Google Brotli decoder out-of-bounds write, CVE-2026-111497 (CVSS 9.2 under v4.0 / 9.8 under v3, CWE-787/CWE-252), hitting anything that decompresses untrusted Brotli streams — Node.js, PHP, Bun, Envoy/Istio, .NET, Python, Java, Ruby — with exploit attempts already logged from the US and Netherlands; and an Adobe Commerce/Magento unrestricted-upload flaw nicknamed "PolyShell," CVE-2026-48356 (CVSS 9.3–9.6), which Sansec says has quietly breached hundreds of stores since mid-2025. Patches: Node.js 22.23.4/24.22.1/26.12.1, PHP 8.5.12-1/8.4.27-1/8.3.36-1/8.2.35-1, Microsoft's out-of-band .NET 8.0.33/9.0.22/10.0.14 release. The Hacker News (source) · CISA (source)
🟠 High impact

CVE-2026-97631 — 16-year-old UEFI/EDK II PXE flaw still has no patch

CVE-2026-97631EDK IICVSS 9.5no patch

WhatA 16-year-old PXE boot flaw in EDK II runs before Secure Boot and can give a network attacker code execution on network-booting servers.

DoDisable PXE boot until a stable firmware release ships the upstream fix.

Details and sources
A flaw tracing to a November 2010 commit in TianoCore EDK II's PXE network-boot code, CVE-2026-97631 (CVSS 9.5), runs before Secure Boot engages and can let a network attacker get code execution on network-booting servers. Affects RHEL 8/9/10, SUSE and Debian's edk2 packages; Intel and Arm say their current products aren't affected, while AMD, Microsoft, AMI, Phoenix and Amazon hadn't responded as of publication. The fix merged upstream Oct 15, but no stable firmware release ships it yet — the only mitigation is disabling PXE boot. No known exploitation so far. The Hacker News (source)
🔴 Exploited or active

CVE-2026-86950 — Apple backports exploited CoreGraphics fix to older devices

CVE-2026-86950Apple CoreGraphicsexploited

WhatApple backported its fix for an exploited CoreGraphics out-of-bounds write, used in targeted attacks, to older iPhones and iPads.

DoUpdate older devices to iOS 18.7.11 or iPadOS 18.7.11.

Details and sources
Apple backported its fix for an already-exploited CoreGraphics out-of-bounds-write flaw, CVE-2026-86950 (CVSS 8.8, originally patched in September), to iOS 18.7.11/iPadOS 18.7.11 for older hardware (iPhone XS/XR, several older iPads) on Oct 26–27. Apple says it's aware of a report the bug "may have been exploited in an extremely sophisticated attack against specific targeted individuals" — narrow, targeted exploitation, not mass. The Hacker News (source)
🟠 High impact

CVE-2026-80097 — Microsoft Authenticator (Android) token theft

CVE-2026-80097Microsoft Authenticator AndroidCVSS 8.6

WhatAny installed Android app without permissions could steal work-account authentication tokens from Microsoft Authenticator; no known exploitation.

DoUpdate Microsoft Authenticator for Android to version 6.2606.4246.

Details and sources
A flaw, CVE-2026-80097 (CVSS 8.6), let any installed, permissionless Android app steal work-account authentication tokens from Microsoft Authenticator. Fixed in version 6.2606.4246; no known in-the-wild exploitation. The Hacker News (source)
🔴 Exploited or active

PHP2Shell exploitation expands — Fragnesia kernel flaw chained to root firewalls, PAN-OS also targeted

CVE-2026-109618CVE-2026-109619CVE-2026-46300in KEV

WhatA suspected China-nexus actor chained PHP2Shell with the Fragnesia kernel bug to root 25 firewalls, and attackers also hit PAN-OS management interfaces.

DoPatch PHP to 8.5.12/8.4.27/8.3.36/8.2.35, and patch the kernel for Fragnesia before the Oct 31 KEV deadline.

Details and sources
Researchers at Dream found a suspected China-nexus actor chaining the PHP2Shell flaws (CVE-2026-109618/CVE-2026-109619, CVSS 9.2 each, patched Oct 21 in PHP 8.5.12/8.4.27/8.3.36/8.2.35) with a Linux kernel privilege-escalation bug nicknamed "Fragnesia" (CVE-2026-46300, CVSS 7.8) to root 25 internet-facing firewalls across the US, Japan, Taiwan and South Korea — including three in a US government-affiliated environment where the attacker went after Active Directory. The attacker's tooling reportedly shows traces of Anthropic's Claude Code and an AI-driven offensive tool built on DeepSeek/Kimi models. Separately, Palo Alto Networks confirmed threat actors are exploiting the same PHP2Shell flaws against internet-exposed PAN-OS management interfaces. CISA added CVE-2026-46300 to its KEV catalog, with a federal patch deadline of Oct 31. The Hacker News (source) · The Hacker News (source)
🔴 Exploited or active

CVE-2026-76418 — actively exploited Cisco ASA/FTD SSL VPN DoS flaw

Added after publishingCVE-2026-76418Cisco ASA/FTD SSL VPNexploited

WhatAn unauthenticated crafted request during a VPN connection attempt forces a device reload when secondary authentication is configured; Cisco confirms active exploitation.

DoInstall fixed ASA 9.16 to 9.24 or FTD 7.0 to 10.1 releases, or disable secondary authentication on affected tunnel groups meanwhile.

Details and sources
An unauthenticated, remote flaw, CVE-2026-76418 (CVSS 8.6), in Cisco ASA/FTD's Remote Access SSL VPN service lets an attacker send a crafted request during a VPN connection attempt — no successful login required — to force a device reload, when secondary authentication is configured on a tunnel group. Cisco says it became aware of active exploitation in October 2026 but hasn't disclosed scale, timing, or attribution. Fixed releases span ASA 9.16–9.24 and FTD 7.0–10.1; where patching isn't immediate, Cisco recommends disabling secondary authentication on affected tunnel groups or the default group as a temporary mitigation. The Hacker News (source)

Breaches / Incidents 6

🟠 High impact

University of New Mexico Health System — cyberattack diverts ambulances

UNM Health Systemhealthcare"BYOD" claim

WhatA cyberattack on Oct 26 forced New Mexico's only Level 1 trauma hospital to divert ambulances and cancel surgeries; a group called BYOD claims data theft.

Details and sources
A cyberattack on Oct 26 forced New Mexico's only Level 1 trauma/burn/children's hospital complex to divert ambulances for about two hours and cancel outpatient surgeries and classes at its medicine, nursing and pharmacy colleges. The university hasn't confirmed or denied ransomware; a little-known group calling itself "BYOD" claimed it stole data from 36,000 machines — unconfirmed. The Record (source)
🟠 High impact

GMO Internet (Japan) FX-trading cloud breach escalates

GMO InternetJapanFX trading cloud

WhatGMO refused a roughly $50 million extortion demand after a breach of its Desktop Cloud service; 26,739 servers potentially affected and OANDA Japan saw unauthorized trades.

Details and sources
GMO Internet Group is offering a $1 million bounty for information leading to an arrest, after refusing a roughly $50 million extortion demand tied to a breach of its Onamae.com Desktop Cloud service (hosts Windows desktops used for FX trading). 26,739 servers are potentially affected; OANDA Japan separately confirmed unauthorized trades using credentials believed stolen from GMO's servers. No attribution yet; no ransomware deployed so far. The Hacker News (source)
🟠 High impact

Keystone hardware wallet maker — device-authenticity key and customer data stolen

Keystonehardware walletkey stolen

WhatAn attacker abused a third-party monitoring tool flaw to steal about 26,000 customer records and the key Keystone devices use to prove authenticity.

DoKeystone will destroy 18,835 unsold devices carrying the compromised key.

Details and sources
An attacker used a flaw in a third-party monitoring tool to escalate privileges on Keystone's website, stealing roughly 26,000 customer records (names, addresses, order details — no recovery phrases or payment data) and the cryptographic key its devices use to prove authenticity. Keystone will destroy 18,835 unsold devices carrying the compromised key. The Hacker News (source)
🟡 Worth knowing

UK policing watchdog: suppliers holding police data repeatedly breached since 2021

UK policingthird-party suppliersHMICFRS

WhatA watchdog report finds suppliers holding British police data were repeatedly compromised since 2021, without confirming data theft; 252 incidents logged in Q1 2026.

Details and sources
HMICFRS's new report finds hackers have repeatedly compromised third-party suppliers that hold British police data, though it stops short of confirming data was actually stolen in those incidents. Separately, the Police Digital Service logged 252 cybersecurity incidents and 796 "near misses" across UK policing in Q1 2026 alone. The Record (source)
🟠 High impact

DHS "Icepick" leak exposes 41,000+ employee and contractor records

DHS"Icepick" leakOkta claim unconfirmed

WhatDDoSecrets published over 41,000 DHS employee and contractor records, attributed to DHS's Okta login; Okta says it has no evidence of compromise.

Details and sources
Distributed Denial of Secrets published an archive — dubbed "Icepick" — of more than 41,000 DHS employee and contractor records spanning CBP (~21,000), ICE (~10,000), USCIS (~6,200) and CISA (~1,000), including cellphone numbers, office locations, ICE/CBP facility addresses, and internal group names tied to wiretap programs, Palantir software and xAI's Grok chatbot — some suggesting undercover/surveillance roles. DDoSecrets attributes the source to DHS's Okta login system; Okta says it has no evidence of a compromise, so the claim is unconfirmed. VisaVerge (source)
🟠 High impact

Attacker probed Alpine Linux's own build infrastructure via a critical regex flaw

Added after publishingCVE-2026-97229Alpine LinuxCVSS 10.0

WhatAn attacker sent crafted CI configurations to Alpine's GitLab trying to exploit an Onigmo regex overflow; success is unproven but Alpine treated the server as compromised.

DoAlpine rebuilt the server, reset all account passwords and rotated secret keys.

Details and sources
A newly registered account sent crafted CI configurations against gitlab.alpinelinux.org (which hosts the aports package-build repo) on Oct 20–21, attempting to exploit CVE-2026-97229 (CVSS 10.0), a regex integer-overflow in Onigmo (Ruby's built-in regex engine) — GitLab's own patch covered its bundled Ruby, but Alpine's standard Ruby remained exposed even after an Oct 22 GitLab upgrade. The attempts crashed GitLab worker processes eight times, and the attacker downloaded Alpine's GitLab container image, apparently to tune the exploit. Alpine blocked the account Oct 23, found the server still vulnerable Oct 26, then took it offline Oct 27, rebuilt it, reset all account passwords and rotated secret keys. No evidence the exploit succeeded, but Alpine couldn't rule it out and treated the server as compromised as a precaution. No attribution given. The Hacker News (source)

Malware / Ransomware 4

🟡 Worth knowing

Unnamed ClickFix campaign abuses a fake Cloudflare check

ClickFixfake Cloudflare checkmacOS and Windows

WhatA new unnamed ClickFix campaign uses a fake Cloudflare verification page on a compromised site to make users paste a malicious command.

Details and sources
SANS ISC flagged a new, not-yet-named ClickFix social-engineering campaign using a fake Cloudflare "verification" page on a compromised legitimate website to trick both macOS and Windows users into pasting a malicious command. Both payload-fetch URLs were already dead (403 / non-resolving) by the time of analysis, so the actual second-stage malware was never recovered. SANS ISC (source)
🔴 Exploited or active

"KeyDrift" malware deployed via Cisco Catalyst SD-WAN flaw

Added after publishingCVE-2026-20127Cisco SD-WANKeyDriftCVSS 10.0

WhatUAT-11473 exploited a Cisco Catalyst SD-WAN auth bypass to plant KeyDrift malware that collects AES-256 key material and runs an in-memory backdoor.

DoUpgrade to fixed releases such as 20.9.8.2, 20.12.6.1, 20.15.4.2 or 20.18.2.1 and deploy the Talos Snort rules; no workaround exists.

Details and sources
Cisco Talos reports a group it tracks as UAT-11473 exploited CVE-2026-20127 (CVSS 10.0, unauthenticated auth bypass to admin privileges) in Cisco Catalyst SD-WAN Controller/Manager/Validator to plant "KeyDrift" malware — its LittleFilch component hooks the Controller's vdaemon process to collect AES-256 key material passed between routers, while PolyFlame runs in memory as a backdoor/relay. Talos assesses with high confidence that UAT-11473's infrastructure "overlaps considerably" with the Silent Train network covered above, though it stops short of calling them the same group as Volt Typhoon. Targets include space/satellite services, defense/aerospace and policy-research organizations. No workaround exists; fixed releases include 20.9.8.2, 20.12.6.1, 20.15.4.2 and 20.18.2.1. Talos published Snort detection rules Oct 28. The Hacker News (source)
⚪ Context

GitHub launches an install-time "firewall" for npm

Added after publishingGitHubnpmpackage firewall preview

WhatGitHub previewed an npm package firewall that blocks installs by malware advisories, severity, missing attestation, new install scripts and a cooldown period.

Details and sources
GitHub rolled out a public preview of a package firewall sitting between developers/CI and the public npm registry, blocking installs by known-malware advisories, vulnerability severity threshold, missing SLSA attestation where an earlier version had one, newly added pre/post-install scripts, and a configurable cooldown period (1–365 days, default 1) on newly published versions — aimed at malicious packages that run automatically when AI agents or CI runners install dependencies without human review. Available in public preview to GitHub Enterprise Cloud customers; npm-only for now. The Hacker News (source)
🟡 Worth knowing

GhostAction GitHub supply-chain campaign fizzles after GitHub blocks it

GhostActionGitHub Actionssupply chain

WhatThe GhostAction campaign hijacked about 1,900 GitHub accounts to push credential-stealing workflows, but its October waves extracted no secrets after GitHub blocked it.

Details and sources
The GhostAction campaign — which hijacked roughly 1,900 GitHub accounts to push credential-stealing Actions workflows (disguised under innocuous names like "security-audit.yml") across 30,000+ repositories, hunting AWS, npm, PyPI, DockerHub and AI-provider (OpenAI/Anthropic) tokens — had its most recent Oct 10 and 12 waves extract no secrets at all after GitHub began blocking the malicious workflow on Oct 9. The Hacker News (source)

Threat Actor Activity 8

🔴 Exploited or active

ShinyHunters (UNC6240) PeopleSoft campaign turns out to be a 3-bug chain

ShinyHuntersUNC6240CVE-2026-35273PeopleSoft

WhatShinyHunters exploited a three-flaw PeopleSoft chain; a September wave hit dozens of systems across tech, healthcare, agriculture, transport and government.

DoApply Oracle's June 2026 PeopleSoft patch.

Details and sources
Mandiant/Google Threat Intelligence Group attributes an Oracle PeopleSoft exploitation campaign, CVE-2026-35273 (CVSS 9.8, patched June 2026), to ShinyHunters (tracked as UNC6240). Trend Micro's ZDI found the "single" CVE is actually three chained flaws (an SSRF in the Integration Gateway plus a loopback-trust bypass in the Environment Management Hub); a September wave hit "dozens of systems worldwide" across tech, healthcare, agriculture, transport and government, beyond the original May–June wave against universities. The Hacker News (source)
🟠 High impact

"CyberStrikeAI" — 44-agent AI hacking platform hits Japan, Taiwan, Hong Kong

CyberStrikeAIJapan, Taiwan, Hong Kong

WhatA pentesting tool reconfigured with 44 AI agents validated at least 19 targets in October, mostly in Japan, including auto-generated phishing pages.

Details and sources
Trend Micro's TrendAI researchers found an open-source pentesting tool reconfigured with 44 discrete AI agents covering recon through auto-reporting, including auto-generated Japanese-language phishing pages; it validated at least 19 targets in a short October burst, 16+ of them in Japan. No operator attribution yet. The Hacker News (source)
🔴 Exploited or active

"Skyw4lker" device-code phishing campaign targets US manufacturing, tech, healthcare

Skyw4lkerdevice-code phishingMicrosoft 365

WhatAn unattributed campaign uses fake contract documents to get victims to complete a real Microsoft 365 device authorization, then steals the resulting access token.

Details and sources
An unattributed campaign lures victims into completing a genuine Microsoft 365 device-authorization flow via a fake contract document, then steals the resulting real access token rather than phishing a fake login page. The Hacker News (source)
🟡 Worth knowing

FBI seizes "Virtualine" bulletproof-hosting domain

Virtualinebulletproof hostingFBI seizure

WhatUS and European law enforcement seized the domain of bulletproof host Virtualine, which had on-chain links to ransomware and infostealer operators.

Details and sources
DOJ/FBI, with German, Dutch and UK (National Crime Agency) law enforcement, seized the domain of bulletproof host Virtualine (aka Secury, Massgod, Spamhouse, Lanedo, Omegatech), alleged to have hosted child-sexual-abuse material and laundered proceeds via Bitcoin; Chainalysis found on-chain links to ransomware and infostealer operators among its customers. The Hacker News (source)
🟡 Worth knowing

Nefilim/Hive ransomware affiliate sentenced to nearly 6 years

NefilimHivesentencing

WhatUkrainian Artem Stryzhak was sentenced to 70 months for acting as a Nefilim and later Hive ransomware affiliate; a co-conspirator remains at large.

Details and sources
Ukrainian national Artem Stryzhak was sentenced to 70 months for his role as a Nefilim affiliate (2021, operating from Barcelona) and later a Hive affiliate; Hive's ransomware reportedly hit thousands of victims in 80 countries during COVID. Co-conspirator Volodymyr Tymoshchuk remains at large with an $11 million DOJ reward. The Record (source)
🟡 Worth knowing

US indicts six over Madagascar-based romance/fake-NFT scam ring

Madagascarromance and crypto scamindictment

WhatSix people were indicted over a romance and crypto investment scam run from Madagascar that stole at least $516,000 from US victims.

Details and sources
Malaysian national Chui Kok Wa and five co-defendants were indicted over a romance-scam-to-crypto-investment-fraud operation run from Antananarivo, Madagascar (relocated there after an earlier Cambodia crackdown), which stole at least $516,000 from US victims. The Record (source)
🟠 High impact

Volt Typhoon's "Silent Train" covert access network detailed

Volt TyphoonSilent Traincovert network

WhatBlack Lotus Labs mapped Volt Typhoon's Silent Train network, tied to four zero-day campaigns and hundreds of victims; the network has since been disrupted.

Details and sources
Lumen's Black Lotus Labs mapped a covert access network it calls Silent Train, run by China-linked Volt Typhoon since at least 2021 and tied to four zero-day campaigns (Versa Director CVE-2024-39717, N-able N-central, Cisco Catalyst SD-WAN CVE-2026-20127, Arista VeloCloud CVE-2026-16812) plus a likely Quest KACE zero-day affecting 30+ victims. Hundreds of victims span telecom, government, defense contractors, satellite/mission-critical comms, universities and critical infrastructure, with Hawaii, California and Texas the main US focus areas; the network has since been disrupted. The Hacker News (source)
🔴 Exploited or active

Warp Panda (China-nexus) exploited a VMware ESX logging-bypass flaw as a zero-day

Warp PandaCVE-2026-41709VMware ESXin KEV

WhatWarp Panda exploited a VMware ESX logging-bypass flaw as a zero-day to reduce audit visibility before deploying BRICKSTORM and stealing Microsoft 365 tokens.

DoApply the July 2026 VMware ESX patch; federal KEV deadline is Nov 11.

Details and sources
CrowdStrike attributes early-2026 zero-day exploitation of an insufficient-logging flaw in Broadcom VMware ESX (CVE-2026-41709, CVSS 2.7, patched July 2026) to Warp Panda (aka UNC5221/Armory Castle/Clay Typhoon/VerdantBamboo), used to reduce audit-trail visibility before deploying BRICKSTORM and SharpStorm for persistence and harvesting Microsoft 365 session tokens. Targets are mostly US organizations, with the legal sector highlighted. CISA added the CVE to its KEV catalog Oct 28, with a Nov 11 federal deadline. The Hacker News (source)
‹ 28/10/202630/10/2026 ›