Microsoft published CVE-2026-124194 on 10 November in the November Office security update 1, and CISA added it to the known exploited catalog the same day 2. CISA’s entry carries a due date of 24 November 3. The base score is 6.5 and Microsoft rates it Moderate 1, which on a release of 1,093 fixes makes this the lowest-rated item in the batch 4, and the exploitability assessment on the same record reads Exploitation Detected 1.

A 6.5 buys a slot in the middle of a normal patch cycle. The exploitation flag sits on the same page as the score, and the deadline sits on the catalog entry; together they move this one to the front of my queue.

The unsafe action is opening a file

The weakness class is CWE-356, a product interface that does not warn the user before an unsafe action 1. Here the unsafe action is opening a message file. An unauthenticated attacker sends a crafted message file and convinces someone to open it, and Outlook connects to a network location the attacker controls without displaying the usual external-content warning 1. That connection discloses the user’s NTLMv2 hash, which Microsoft says an attacker could use to authenticate as that user 1. The preview pane is not an attack vector; the file has to be opened 1.

Every layer above that click belongs to the mail path I work on. The release notes for 10 November list this CVE in the Outlook section, one of twenty Outlook fixes in the batch 5. If a message carrying the malicious file reached a mailbox, then the filters and the attachment rules each either passed it or missed it, and the final control in the chain is a dialog box inside a desktop application.

I can tighten a mail rule in an afternoon and see the effect in the logs the same day. A warning the client is supposed to raise is somebody else’s code path, shipped on a monthly cadence, and in this case it did not raise.

What the 6.5 measures and what it leaves out

The vector is AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:N 1. Confidentiality is the only impact: nothing executes, nothing is modified, availability is untouched. For one interaction that leaks a user’s credential material, the math holds, and I would not argue for a higher number.

The vector leaves out what that credential is worth here, because the answer depends on which systems still accept it. I treat this as a credential exposure question filed under spoofing, and I look at the places that authenticate against leaked material rather than the places that would run code. The advisory describes the mechanism and leaves that inventory to me 1.

What I check at the gateway

When the client-side warning is the broken layer, the gateway work is about shrinking what the leak is worth.

I start with the attachment type. Message files carry calendar invites and saved mail, so a blanket block costs users real function. The first question is whether a message file can enter the tenant, which sender populations may send one, and whether the gateway rewrites the type on the way in.

Then I check whether any policy suppresses the external-content warning. The bug removed a warning that should have appeared; a configuration that turns warnings off removes it for every message, all the time, and users will not report a prompt they have never seen.

The outbound side matters more here than the inbound side. A desktop that can reach an arbitrary host on the internet is the channel that carries the hash out, and outbound SMB to the internet is the path worth closing at the perimeter. Little legitimate traffic needs it.

Last is NTLM itself. I want a count of systems that still answer an NTLM challenge and the sources those attempts arrive from, because that count decides whether a sniffed hash turns into a session. That work predates this CVE and will outlive it; this one makes the argument for it with a date attached.

On the endpoint side, which is my coverage area, I want telemetry for an Office process opening a connection to an external host, because the disclosure happens after the message is closed and quarantined.

Installed is not the same as fixed

The fix ships in the Office security update, and the advisory’s update table lists Office products only 5. A host that reports current on Windows can be running vulnerable Outlook, and on this bug that combination looks patched from the wrong console.

The 10 November release notes give one build per channel, with the twenty Outlook fixes listed under the same release 5.

Channel Build
Current Channel, Version 2610 20527.20196
Monthly Enterprise, Version 2610 20527.20194
Monthly Enterprise, Version 2609 20430.20264
Monthly Enterprise, Version 2608 20326.20320
Semi-Annual Enterprise, Version 2610 20527.20192
Semi-Annual Enterprise, Version 2609 20430.20262
Semi-Annual Enterprise, Version 2608 20326.20318
Office LTSC 2024, volume licensed 17932.21140
Office LTSC 2021, volume licensed 14334.21048

The advisory names the two LTSC builds and sends Microsoft 365 Apps to the release page without a build number 1. That is why I ask for a number per channel instead of a yes or no. “Office is patched” is a statement about a machine, not about a build.

The order I put in front of the teams

My work on this stops at the recommendation. The deployment side is automated, and the teams that own deployment run it on their own schedule. I add the sequence and the reasoning to the cycle, starting with the build check, because that is the item that closes the bug. The outbound path comes next, because it limits what a leaked hash can reach. The 24 November date anchors both, since it is public and it does not move.

The 6.5 is fair, and I will keep it in the writeup. The reason this one gets a post is that a moderate can carry more risk than a critical on the same day, and a severity-ranked queue only catches that when someone reads the exploitation field.


  1. Microsoft Security Response Center. (2026, November 10). CVE-2026-124194: Microsoft Outlook spoofing vulnerability. Microsoft. https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-124194 ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎

  2. Cybersecurity and Infrastructure Security Agency. (2026, November 10). CISA adds six known exploited vulnerabilities to catalog. CISA. https://www.cisa.gov/news-events/alerts/2026/11/10/cisa-adds-six-known-exploited-vulnerabilities-catalog ↩︎

  3. National Institute of Standards and Technology. (2026, November 10). CVE-2026-124194 detail. NVD. https://nvd.nist.gov/vuln/detail/CVE-2026-124194 ↩︎

  4. Tenable Research Special Operations. (2026, November 10). Microsoft’s November 2026 Patch Tuesday addresses 1093 CVEs (CVE-2026-124194). Tenable. https://www.tenable.com/blog/microsoft-november-2026-patch-tuesday-addresses-1093-cves-cve-2026-124194 ↩︎

  5. Microsoft. (2026, November 10). Release notes for Microsoft Office security updates. Microsoft Learn. https://learn.microsoft.com/en-us/officeupdates/microsoft365-apps-security-updates ↩︎ ↩︎ ↩︎