What happened

Confirmed Barracuda’s advisory covers an authorization bypass in Email Security Gateway only (CVE-2026-116606, CVSS 9.3) and a command injection in ESG, Load Balancer ADC, Message Archiver, WAF Control Center, WAF and Web Security Gateway (CVE-2026-111560, CVSS 8.6). Chained, they give root. On 9 November at 21:00 EST Barracuda added that it had seen exploitation “affecting a small number of organizations”. Barracuda CISA added both to KEV on 10 November, due 13 November, with forensic triage “Yes”, ransomware use “Unknown” and references to Mandiant disclosures MNDT-2026-0041/-0042. CISA KEV

Reported Previdian saw 17 attempts against CVE-2026-116606 from 15 IP addresses in three countries between 9 and 11 November; the IPs weren’t published and the actors are unidentified. No ransomware link has been reported. SecurityWeek Previdian

Who is affected and how it works

Confirmed On 5 November Barracuda changed CVE-2026-116606’s “Authentication Required” field to “No (Administrative UI Access Required)”: no credentials, but the admin UI must be reachable. Barracuda KEV says the bypass lets an attacker invoke config-import and restore a crafted backup, and the injection runs root commands “when the crafted timezone value is subsequently processed”. CISA KEV MAG212 analysis That points to a crafted backup that sets a malicious timezone value. Unknown Tooling, actors, start date and the Mandiant pages’ contents.

What to look for

MAG212 analysis With no published IOCs: config-import or backup-restore events you didn’t start, unexpected timezone changes, and admin-UI connections from outside your management networks.

Investigate, contain and remediate (in order)

1. Restrict the admin UI now to trusted networks; this is Barracuda’s workaround. Barracuda
2. Confirm the latest firmware and the patch ID on every device. Devices several versions behind may need more than one update; apply manually if auto-patching didn’t run. Barracuda Campus
3. Triage before you trust a patched box. MAG212 analysis Treat any appliance with an internet-reachable admin UI since 2 November as possibly compromised: preserve logs, review config history, involve Barracuda support or your IR provider.
4. MAG212 analysis If compromise is suspected, rotate LDAP bind, SMTP relay and admin credentials, and consider rebuilding rather than cleaning.

Verify recovery: the patch ID shows as applied on every device; the admin UI doesn’t answer from an external vantage point; config history re-reviewed after patching. Lesson MAG212 analysis: MSPs should inventory admin-UI exposure across every client appliance this week.

What happened

Confirmed Of 1,342 entries in Microsoft’s November data, 1,094 were first published on 10 November: 1,093 assigned by Microsoft and one by VulDB. By MAG212’s count of that data, 63 Microsoft-assigned CVEs are Critical. Exactly one is marked exploited and none publicly disclosed. MSRC Reported SecurityWeek also counts 1,094; ZDI says “about 1,090” and calls the count “a bit confusing”. SecurityWeek notes 165 republished third-party CVEs (none from Chrome) and that this is the last month of updates for Windows 11 23H2 Enterprise and Education. ZDI counts 30 wormable CVEs by its own definition. SecurityWeek ZDI

The Outlook zero-day: who is affected and how it works

Confirmed CVE-2026-124194 is a spoofing flaw (missing warning before an unsafe action) in Microsoft 365 Apps for Enterprise and Office LTSC 2021/2024, rated “Exploitation Detected”, customer action required. A user must open a crafted message file; Outlook then connects to an attacker-controlled location without the usual warning and discloses the user’s NTLMv2 hash. The Preview Pane is not an attack vector. Credited to “nono” and Benoît Sevens. MSRC KEV: due 24 November, forensic triage “No”. CISA KEV Reported ZDI says the hash could be used “to authenticate as the user”. ZDI Unknown Who exploited it, against whom, the delivery method and IOCs.

The criticals to patch next (per ZDI; none reported exploited)

Look for, contain and remediate (in order)

1. Deploy the November Office updates to Microsoft 365 Apps and Office LTSC 2021/2024. MSRC
2. Patch Exchange servers and domain controllers first for CVE-2026-116928 and CVE-2026-119958. ZDI MAG212 analysis Zero-click and unauthenticated, so expect exploit development.
3. MAG212 analysis Block outbound SMB/NTLM (TCP 445) to the internet and consider NTLM restrictions; review firewall logs for workstations, particularly Outlook, reaching internet hosts on 445.
4. MAG212 analysis Reset passwords for anyone who reports opening an odd message file and review their sign-ins.
5. Plan Windows 11 23H2 Enterprise/Education migrations. SecurityWeek

Verify recovery: patch reporting shows the November Office build on all Outlook clients and the November update on Exchange and DCs; outbound TCP 445 to the internet is blocked from inside. Lesson MAG212 analysis: this month’s only exploited bug is “Moderate”. Exploitation status should outrank CVSS, which is what KEV encodes.

What happened

Confirmed Vates rates VSA-2026-062, “Empty-object login token”, Critical and says it is being actively exploited by an actor operating initially from 91.211.251.112:8001 and an onion address. There is no CVE yet. Vates Reported Credited to Filip Magic and Christoffer Olsen (TDC NET Security); THN adds that VirusTotal flags that IP as a RedLine Stealer C2. A crafted cookie, authenticationToken=j:%7B%7D, is parsed into an empty object that the token provider accepts, giving XO admin, then code execution on pools and root on hosts. The Hacker News Unknown Start date and victim count.

Who is affected

What to look for (vendor IOCs)

# VMs whose platform settings include a file-backed serial port (IOC: hvm_serial: file)
xe vm-list params=name-label,platform | grep -B1 'hvm_serial: file'
# Search current and rotated xensource logs for file-backed serial configuration
zgrep -- '-serial file:' /var/log/xensource.log*
# Preload hijack and the published persistence files
ls -la /etc/ld.so.preload /.ustar* /etc/cron.d/zz-ustar*

Contain and remediate (in order, per Vates)

1. Take XO off the internet onto a dedicated management network; block the IPs and onion address.
2. Upgrade to XO 6.9.2 or 6.10.1 (xo-server 5.211.2 / 5.213.1); from-sources builds to latest master.
3. Optional interim WAF or reverse-proxy rule for the token patterns; Vates warns regex rules can be bypassed.
4. Run the host checks above on every host.
5. If compromised: rotate all secrets (proxy, backup remotes, root, pool secret, users, OTP, mail) and fully reinstall. Vates

Verify recovery: confirm the XO version, that the interface is unreachable from untrusted networks, re-run the checks, and confirm rotated secrets are in use. MAG212 analysis With no logs, reinstalling is the only high-confidence recovery for a host with any hit. Virtualisation control planes belong on an isolated network, especially for MSPs and schools that chose XCP-ng to save cost.

Chrome CVE-2026-125350: exploited XML use-after-free

Reported Google is “aware that an exploit for CVE-2026-125350 exists in the wild”. Reported anonymously on 6 November, it is the eighth exploited Chrome zero-day of 2026, fixed in Chrome 157.0.8103.50/.51 (Windows, macOS) and 157.0.8103.50 (Linux). The Hacker News SecurityWeek Confirmed KEV notes it affects other Chromium browsers such as Edge and Opera. CISA KEV Unknown Attack details, targets, IOCs, and whether Edge has shipped a fix; Google’s release note was not opened. Do MAG212 analysis push 157.0.8103.50/.51 and force a browser restart, then track Edge and Opera. Verify the version in your endpoint inventory.

Cisco BroadWorks CVE-2026-20320: exploited XXE

Confirmed An XML external entity flaw that “could allow an unauthenticated, remote attacker to read sensitive configuration information”; due 24 November. CISA KEV Unknown Fixed releases, CVSS and IOCs; read Cisco’s advisory (not opened) before acting. MAG212 analysis Configuration can hold credentials, so plan rotation alongside the patch.

Openwave Webtop Media Transcoder CVE-2026-111793

Confirmed “Incorrect resource transfer between spheres” in Openwave Messaging Webtop Media Transcoder; due 24 November, referencing the vendor’s GitHub advisory and JVN. CISA KEV Openwave Unknown Versions and exploitation details. Narrow audience: telecom and messaging providers.

MAG212 analysis Where we recorded the flag, every three-day KEV deadline this week carries forensic triage “Yes” and every 14-day one “No”: the short clock appears to signal suspected compromise, not just urgency.

What happened

Confirmed The FortiGuard PSIRT index grew from 324 to 370 advisories on 10 November. Every advisory page our researchers fetched says “Known Exploited: No” and offers no workaround, so upgrading is the only fix. FortiGuard Fortinet RSS Reported The fixes came from Fortinet’s AI-assisted testing programme; watchTowr’s Yordan Ganchev expects them to be “rapidly reverse-engineered, reproduced, and exploited” and advises restricting exposed management interfaces. Cybernews

Contain and remediate (in order)

1. MAG212 analysis Start with FortiManager/FortiAnalyzer reachable on FGFM from untrusted networks and internet-facing FortiMail: unauthenticated code execution, the class watchTowr expects to be weaponised.
2. Schedule migrations now for FortiMail 7.2 and FortiManager/FortiAnalyzer 7.4/7.2, which have no in-branch fix. FG-IR-26-199 FG-IR-26-275
3. Upgrade FortiOS for the OSPF flaw. MAG212 analysis Meanwhile, OSPF authentication and passive interfaces on user segments reduce adjacency exposure (general hardening, not Fortinet guidance).
4. Read FG-IR-26-271 and -265 for FortiAuthenticator versions and patch; it sits in front of VPN and admin MFA.
5. Restrict all management interfaces to trusted networks. Cybernews

Look for and verify: MAG212 analysis no IOCs exist; watch FGFM and FortiMail HA traffic from unexpected sources and OSPF neighbour changes you didn’t make. Confirm firmware per device and, from outside, that FGFM and FortiMail admin/HA ports don’t answer.

What happened

Confirmed After a year-long investigation, Check Point Research linked three clusters, SilentDelivery, OceanGate and DarkWire, through a shared framework it calls Sparrow, which it suggests is supplied by a developer it names CopperDragon. Targets: telecom providers, government, military, research and critical infrastructure in 21 countries. CPR warns a telecom compromise could expose call records, subscriber locations, SMS and possibly lawful-intercept systems. Check Point Check Point Research

How each cluster gets in

• SilentDelivery (Europe/Asia, 2026): obfuscated JavaScript on legitimate sites fingerprints visitors and shows a fake Chrome certificate warning offering a “Certificate Fix”, a trojanized MSI that drops a backdoor. Also LNK files disguised as PDFs sent to Indonesian organisations. The Hacker News
• OceanGate (2025–26, India focus): a “vulnerable internet-facing Linux server”, with no product or CVE named, then RDP brute force to an admin account; Linux backdoors with an optional rootkit, Windows keylogger and credential stealers. The Hacker News
• DarkWire (mid-2025, Latin American telecom and government): DLL sideloading, DNS tunnelling over MX queries, C2 often in the victim’s own country. The Hacker News ESET documented related Sparrow activity in Latin America in September. CSA

Attribution: assessed, not proven

Reported CopperDragon is “assessed” to supply FamousSparrow, Earth Estries, Tropic Trooper, UAT-9244 and UAT-8302; FamousSparrow’s relationship to Salt Typhoon “remains unclear”. The Hacker News Confirmed CPR’s own caveat: shared malware or C2 “does not necessarily mean” the same actor, and it gives no formal confidence level. Check Point Unknown No government corroboration found. We say “shared toolkit links three campaigns”, not “one group”.

What to look for, and what to do

CPR publishes domains, IP addresses and file hashes, plus host artifacts: a Run-key persistence value, a local account added to Administrators with remote token filtering disabled, custom firewall rules, distinctive C2 HTTP paths and a renamed LSASS dump. We don’t reprint the values: copy them verbatim from the Check Point Research report into your tooling. Check Point’s guidance, in order: Check Point

1. Sweep DNS, proxy and EDR telemetry for CPR’s domains, IPs and hashes; hunt the host artifacts.
2. Treat any “certificate fix” or update utility as suspicious. MAG212 analysis Block unsigned MSIs from browser download folders.
3. Monitor for malicious DLL loading, credential dumping, keylogging and new persistence.
4. Restrict and monitor RDP, SMB and SSH, especially brute force on internet-facing hosts.
5. Inspect DNS, including tunnelling over MX, and unusual outbound traffic; inventory and patch internet-facing Linux servers.

Verify: MAG212 analysis after removal, the rogue account and firewall rules are gone and token-filtering policy is back to baseline; rotate credentials on any host with credential-dumping artifacts.

NEAR AI: an exploited SGLang flaw, and a TEE that didn’t help

Confirmed On 8 November an attacker exploited an RCE in SGLang, “the open-source inference engine running on our servers”, reached NEAR AI infrastructure and an internal system behind the deprecated Private Chat service. Exposed: names and emails of about 100 users (some with plan and balance data), chat history, settings and logs of 18 agents on a legacy deployment outside the TEE, and two Cloud customers’ API keys (now revoked) that sat in logs. No sign stored Private Chat conversations were accessed, but the attacker “could also have observed user requests”, including confidential-mode ones. NEAR AI revoked sessions and admin credentials, took Private Chat offline, patched every inference server, rotated credentials and notified users. NEAR AI Reported The TEE “does not protect against a vulnerability in the software running inside the environment”. Brave’s Leo uses a NEAR AI-run TEE; NEAR hasn’t said whether the server handled Brave traffic. The Hacker News

Unknown No CVE, endpoint, SGLang version or IOCs, and no SGLang advisory tied to the incident. CERT/CC listed six unpatched SGLang flaws in July, but nothing links NEAR AI’s breach to any of them. SGLang operators (CERT/CC): keep SGLang off untrusted networks, segment it, set SGLANG_USE_PICKLE_IPC=false (a configuration change; restart the process, MAG212 analysis) and disable unused endpoints. CERT/CC MAG212 analysis Also require API keys and keep secrets out of inference-host logs.

Fireworks AI: an LLMjacking crew came in through a third-party cloud provider

Confirmed An unnamed cloud provider hosting Fireworks R&D machines had an incident on 23 October; on 24 October the attacker reached development machines and ran an unauthorized workload, detected the same day. Accessed: resource metadata, billing and shipping addresses, invoices, balances, secrets for a limited set of accounts and some incomplete model weights and datasets. Not affected: Inference, Training SDK and Nexus; card numbers aren’t stored. Fireworks attributes it to an unnamed, financially motivated group reselling tokens through illicit gateways (LLMjacking), engaged Unit 42 and told law enforcement. Fireworks Reported Affected accounts are a “limited number”, all notified. The Record reports Notion was notified on 2 November, disconnected, disabled its Insights feature and stopped serving open-weight models via Fireworks. The Record We do not report that Notion AI chats were exposed: a THN headline says so, but neither its article body nor Fireworks’ bulletin does, and we haven’t read Notion’s advisory. The Hacker News

Fireworks customers MAG212 analysis, from the bulletin: rotate secrets stored in Fireworks and all API keys; look for keys you didn’t create and odd usage or billing; treat uploaded datasets and fine-tunes as possibly exposed; alert on LLM usage spikes. Lesson: treat AI vendors as holders of your secrets, with scoped, rotatable keys on your third-party risk register.

Confirmed Vanderbilt has an active incident that may affect devices, systems and cloud services, and told users to “SHUT DOWN YOUR DEVICE IMMEDIATELY” and leave university Wi-Fi. It has about 14,000 students and more than 6,000 staff. WSMV The Record Reported Officials called it a “likely” ransomware attack; classes proceeded though some were cancelled; users reported ransom notes and files renamed with a .byod extension (Nashville Scene via Gizmodo, not officially confirmed). Gizmodo Claimed BYOD says it holds data on more than 1 million people including SSNs, plus the domain controller and research data. Vanderbilt didn’t respond to Gizmodo. Gizmodo

Reported BYOD’s first victim appeared on 29 September; it has also claimed Michigan State and UNM Health Sciences Center, which confirmed a “cybersecurity event” on 26 October. Its UNM machine count is unverified, as is a 9 November claim against the Texas Health and Human Services Commission. The Record The Register Unknown Vector, official ransomware confirmation, restoration status and whether data was taken. No IOCs.

For school and university IT MAG212 analysis: (1) confirm tested offline or immutable backups of domain controllers and identity systems; (2) rehearse a campus-wide “shut down devices” message on a channel that works without email; (3) review privileged and domain-admin activity; (4) watch Vanderbilt’s status page for a vector disclosure.

Defender’s Action Desk

Prioritised from today’s stories. MAG212 analysis for the ordering; sources are in each article.

• NowBarracuda: admin UI to trusted networks only; confirm patch IDs; review config-import/restore history; triage exposed boxes before 13 Nov. → 01
• NowXen Orchestra: isolate; XO 6.9.2 / 6.10.1; block 8 IPs + onion; run 3 host checks; any hit → rotate and reinstall. → 03
• NowOutlook and Chrome: November Office update; Chrome 157.0.8103.50/.51 with forced restart. → 02 → 04
• TodayExchange and DCs: first ring for CVE-2026-116928 and CVE-2026-119958; block outbound TCP 445. → 02
• TodayFortinet: close FGFM and FortiMail exposure; book FortiMail 7.2 and FortiManager/FAZ 7.4/7.2 migrations. → 05
• TodayN-able on-prem N-central: keep offline until the hotfix is verified (Briefly noted).
• This weekAI vendors: rotate Fireworks keys and secrets; keep SGLang off untrusted networks. → 07
• This weekHunting: load CPR’s CopperDragon IOCs; brief staff on fake “certificate fixes”. → 06
• This weekGit: update to 2.56.1 or your branch’s fix; no recursive clones of untrusted repos (Briefly noted).

KEV deadlines (federal due dates; a priority signal for everyone)