XHack Logo
XHack
Products
Services
Compliance
Pricing
Resources
Company
Sign upLogin
XHack Logo
XHackOffensive Security

Certified offensive security team delivering penetration testing evidence written for your auditor.

OSCP+OSCPC-AI/MLPenCASA
support@xhack.io

24/7 SOC Operations

XHack Status
Under attack? Get help now
Services
  • VAPT Services
  • Red Teaming
  • SOC Services
  • Threat Intelligence
  • Incident Response
  • Managed Testing
Pricing
  • Platform Plans
  • Services Pricing
Compliance
  • SOC 2
  • PCI DSS
  • ISO 27001
  • GDPR
  • HIPAA
  • ISO 42001
  • AI Maturity Assessment
  • TX-RAMP
  • NBFC / SECP
  • All Frameworks
Products
  • Vulnerability Assessment
  • GitGuard
  • AI Probe
  • SOC Dashboard
  • AI Agent
  • Cloud Investigation
Comparison
  • XBOW vs XHack
  • Horizon3 vs XHack
  • Strix vs XHack
  • Pentera vs XHack
Resources
  • Platform Tour
  • All Features
  • Install the Agent
  • XHack AI
  • Documentation
  • Blog
  • Case Studies
  • Documents
  • FAQ
Company
  • About Us
  • Our Team
  • Certifications
  • Security and Trust
  • VAPT Explained
  • Contact

© 2026 XHACK (SMC-PRIVATE) LIMITED. All rights reserved.

Registered with the SECP, CUIN 0361945. Verify

Security & TrustVulnerability DisclosurePrivacy PolicyTerms of ServiceRefund Policy
Blog/News

CVE-2026-96951: The Third Exploited Windows Shell Flaw of 2026

XHack

XHack

Author

October 26, 2026

12 min read

CVE-2026-96951: The Third Exploited Windows Shell Flaw of 2026

Table of contents

19

What CVE-2026-96951 Actually Is

What Mark of the Web Does, and What CVE-2026-96951 Takes Away

How an Attack Using CVE-2026-96951 Would Reach a User

CVE-2026-96951 Is the Third Windows Shell Bug Exploited This Year

The first two are linked, the third is not yet

What the Public Record Does Not Tell You About CVE-2026-96951

Are You Exposed to CVE-2026-96951?

Fixing CVE-2026-96951

Hunting for CVE-2026-96951 Activity

The Pattern: A Name Search Would Miss These

Where XHack Fits After CVE-2026-96951

FAQ: CVE-2026-96951 Questions Answered

What is CVE-2026-96951?

Is CVE-2026-96951 being exploited?

What does a Mark of the Web bypass actually do?

Which versions does CVE-2026-96951 affect?

Is CVE-2026-96951 linked to APT28?

How do I check for CVE-2026-96951 exposure?

The Bottom Line on CVE-2026-96951

Read this in 30 seconds: CVE-2026-96951 lets a crafted file skip the warnings Windows shows for downloaded files, and Microsoft says attackers were already using it when the patch shipped. It is the third Windows Shell bug exploited in 2026, and nobody has published how it works.

  • It removes the warning, not the user. Microsoft’s record says the flaw lets an attacker “evade Mark of the Web (MOTW) defenses”, and the victim still has to open the file. CVSS 3.1 is 8.8 and Microsoft’s severity is Important.
  • Exploited when it shipped. Microsoft’s record says exploited: Yes, publicly disclosed: No. CISA listed it the same day, October 13, with an October 16 deadline and the forensic triage flag.
  • It is the third in eight months. CVE-2026-21510 came in February, and CVE-2026-32202 in April, which Akamai calls an incomplete patch of the first and Microsoft re-released on July 28.
  • The public record has holes. We found no analysis of the mechanism, no attribution, and sources disagree on who reported it.
  • The fix is a build number. Windows 11 24H2 needs 10.0.26100.9711 or later, and Windows 10 22H2 needs 10.0.19045.7807.

A Mark of the Web bypass does not break into a computer. It removes the one speed bump that makes people hesitate before opening a file.

CVE-2026-96951 is easy to underrate: it needs a person to click, it carries an 8.8, and Microsoft calls it Important. This article covers what the bug removes, how the three 2026 Shell bugs connect, what the public record does not say, and how to check that your builds are fixed.

What CVE-2026-96951 Actually Is

Microsoft published the advisory on October 13, 2026, titled “Windows Shell Security Feature Bypass Vulnerability”. The description says improper neutralization of special elements used in a command, a command injection, in Windows Shell “allows an unauthorized attacker to bypass a security feature over a network.”

We read the record through Microsoft’s update-guide data, NVD and the CISA KEV JSON:

  • Weakness: CWE-77, command injection. Microsoft’s impact label is Security Feature Bypass.
  • Score: CVSS 3.1 8.8, AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H, with a temporal score of 8.2. The scores come from Microsoft. NVD lists the record as Analyzed but carries no score of its own.
  • Exploited: Microsoft’s record says Yes, with the release note “Exploitation Detected”. Publicly disclosed: No.
  • CISA’s tags in the NVD record: exploitation active, automatable no, technical impact total.
  • KEV: added October 13, due October 16, forensic triage Yes, ransomware use Unknown.
  • Revisions: only the original release, so nothing has changed since October 13.

The security feature in question is Mark of the Web. Microsoft’s FAQ says an attacker “could evade Mark of the Web (MOTW) defenses.”

What Mark of the Web Does, and What CVE-2026-96951 Takes Away

Windows tags files that arrive from the internet with a hidden NTFS stream called Zone.Identifier. Eric Lawrence’s long-running explainer on downloads and the Mark of the Web lists what that tag triggers:

  • SmartScreen and antivirus checks when a marked executable is launched through Explorer.
  • A “Downloaded from the Internet” prompt for risky file types.
  • Protected View for Office documents, and macros disabled in documents from the internet.
  • Blocks on risky file types such as .js, .lnk, .bat and .dll under Smart App Control on Windows 11.

A file that loses its tag, or never gets one, skips all of that and behaves like a file you made yourself. That is the whole value of a bypass like CVE-2026-96951: every defense downstream of the tag quietly turns off.

What the Mark of the Web controls and what CVE-2026-96951 evades: SmartScreen checks, the internet prompt, Protected View and blocks on risky file types
One hidden tag switches four defenses on

How an Attack Using CVE-2026-96951 Would Reach a User

Microsoft’s FAQ describes two routes. In an email or instant message attack, the attacker sends the target “a specially crafted file that is designed to exploit the vulnerability.” In any case, Microsoft says, the attacker “would have no way to force a user to view attacker-controlled content”, so they must convince the user to click a link or open an attachment.

Help Net Security adds that the file could arrive through email, instant messages, collaboration tools such as Teams and Slack, or a malicious website. It also says the file opens without Protected View and without a block.

That framing explains the score. UI:R means the user must act, which holds it at 8.8, and it explains CISA’s “automatable: no” tag. But a crafted file sent by email or chat is exactly the lure attackers use, and a bypass like this one makes that file look ordinary.

CVE-2026-96951 Is the Third Windows Shell Bug Exploited This Year

Tenable’s Satnam Narang, quoted by Help Net Security, called CVE-2026-96951 “the third flaw in Windows Shell that was exploited in the wild this year.” We checked each one against Microsoft’s records and the KEV feed.

CVE-2026-21510CVE-2026-32202CVE-2026-96951
Microsoft titleShell Security Feature BypassShell SpoofingShell Security Feature Bypass
ReleasedFeb 10, 2026Apr 14, 2026Oct 13, 2026
CVSS 3.18.84.38.8
WeaknessProtection mechanism failureProtection mechanism failureCommand injection
Publicly disclosedYesNoNo
KEV added, dueFeb 10, Mar 3Apr 28, May 12Oct 13, Oct 16
CVE-2026-96951 timeline of the three exploited Windows Shell bugs of 2026: CVE-2026-21510 in February, CVE-2026-32202 in April with a July re-release, and CVE-2026-96951 in October
Three exploited Windows Shell bugs in 245 days

The gaps are 63 days from the first to the second, 182 days from the second to the third, and 245 days from the first to the third.

The first two are linked, the third is not yet

Akamai’s analysis, published April 23, says CVE-2026-21510 was part of an APT28 chain that bypassed SmartScreen and ran attacker-controlled code, and that Microsoft’s February fix missed a step. Explorer resolved a network path before any trust check ran, so a victim machine could still authenticate to an attacker’s server. That residue became CVE-2026-32202.

Microsoft’s own record shows the follow-through. It corrected CVE-2026-32202’s exploited flag and CVSS vector on April 27, then re-released the CVE on July 28 “to comprehensively address the vulnerability” and recommended the July 2026 updates.

So two of the three form a pair. Whether CVE-2026-96951 belongs to the same family is the open question, and Narang says it is unclear whether it is linked to APT28 or to another incomplete fix.

What the Public Record Does Not Tell You About CVE-2026-96951

This section matters more than usual, because the gaps are large.

  • How it works. Microsoft names command injection and the MOTW outcome. No security vendor has published a root-cause analysis that we could find. Anything about the mechanism beyond that would be a guess, so we are not guessing.
  • Which files. Microsoft’s FAQ says “a specially crafted file” and gives no file type. BleepingComputer says no details have been shared on how attackers exploited the flaw or who disclosed it.
  • Who reported it. The update-guide data we read lists no credits. Born City’s coverage names Fujitsu, Anthropic and an independent researcher as credited reporters for the release, and we could not tie them to this CVE.
  • Who is using it. No attribution has been published.
  • Where it was used. No victim count and no campaign report exist yet.

We also could not verify the claim in one outlet that Windows 10 machines outside the Extended Security Updates program do not receive the fix. It fits the build list below, which covers only supported Windows 10 servicing branches.

Are You Exposed to CVE-2026-96951?

If you run Windows, almost certainly yes until you install the October 13 updates. Born City, citing Microsoft, says all supported Windows client and server versions are affected. NVD’s configuration data also lists Outlook Web.

These are the first fixed builds, taken from NVD’s configuration data:

ProductFixed in build
Windows 11 25H210.0.26200.9711
Windows 11 24H210.0.26100.9711
Windows 11 23H210.0.22631.7652
Windows 11 26H110.0.28000.3231
Windows 11 26H210.0.26300.9711
Windows 10 22H210.0.19045.7807
Windows 10 21H210.0.19044.7807
Windows 10 180910.0.17763.9376
Windows 10 160710.0.14393.9612
Windows Server 202510.0.26100.33650
Windows Server 202210.0.20348.5761
Windows Server 201910.0.17763.9376
Windows Server 201610.0.14393.9612

Windows Server 2012 is also listed as affected without a build number in NVD’s data, so check Microsoft’s advisory for that branch.

CVE-2026-96951 fixed builds for Windows 11, Windows 10 and Windows Server, with the PowerShell check for your build and revision
Check your build against the first fixed one

Your real exposure to CVE-2026-96951 depends on who can receive a file. A server nobody opens attachments on is a lower priority than a laptop that handles email, Teams and Slack all day.

Fixing CVE-2026-96951

CISA flagged the CVE-2026-96951 KEV entry for forensic triage, so patching is the first step and not the last.

  1. Install the October 2026 cumulative update and confirm the build number against the table above. winver shows it as the OS Build, and PowerShell can read it from the registry:
$v = Get-ItemProperty "HKLM:\SOFTWARE\Microsoft\Windows NT\CurrentVersion"
"$($v.CurrentBuild).$($v.UBR)"
  1. Patch endpoints that handle outside files first. Mail clients, collaboration tools and browsers are where the crafted file arrives.
  2. Do not rely on Protected View alone. The bug exists to defeat it.
  3. Filter risky attachment types at the mail gateway. This is our suggestion, not Microsoft’s guidance, and it adds a layer that does not depend on the MOTW tag.
  4. Treat unpatched, exposed users as possibly affected. Microsoft marked exploitation as detected at release, so the window before October 13 matters.

Hunting for CVE-2026-96951 Activity

Microsoft published no indicators and no file hashes for CVE-2026-96951, and we found none elsewhere. Born City reports that Sigma rules and KQL queries exist, and we have not reviewed them.

What a hunter can do is look for the consequence: files from internet sources that arrived without a tag. This is our own coarse heuristic, and it is noisy.

# Files in Downloads with no Zone.Identifier stream (candidates, not proof)
Get-ChildItem "$env:USERPROFILE\Downloads" -File -Recurse |
  Where-Object { -not (Get-Item $_.FullName -Stream Zone.Identifier -ErrorAction SilentlyContinue) } |
  Select-Object FullName, LastWriteTime

Many legitimate files lack the tag. Lawrence’s explainer lists the usual reasons: copies to FAT drives, extraction by tools that do not propagate the mark, and downloads with curl or scripts. Use the result to ask which untagged files arrived by email or chat, and look at those first.

The Pattern: A Name Search Would Miss These

The KEV catalog is easy to search for a technique by name, and that search misleads here. We queried the live JSON: 12 entries are named for Mark of the Web or SmartScreen bypasses, from November 2022 to February 2025. None of the 2026 Shell entries appears under that name. CVE-2026-21510 is filed as “Protection Mechanism Failure”, and CVE-2026-96951 as “Command Injection”.

So a dashboard filtering on “Mark of the Web” shows nothing new for 2026, while three exploited Shell bugs sit outside the filter. Microsoft also accounts for 40 of the 272 entries added to KEV in 2026.

If you track exploited bugs by technique, search the descriptions and Microsoft’s own titles as well. Our post on CVE-2025-61624 covers a similar scoring blind spot, a medium-rated bug that mattered because of the chain around it.

Where XHack Fits After CVE-2026-96951

Quick disclosure: we sell offensive testing, so read this with that in mind.

A penetration test cannot tell you whether a user will open a crafted file, and we will not claim it can. What testing can do is check what a successful file would reach: endpoint controls, attachment filtering and how far one compromised laptop gets. Our AI VAPT services cover that, and managed engagements start at $2,500 on the pricing page. Red teaming is scoped individually.

Our terms say AI output can be incomplete or wrong, so a person reviews anything you would act on. XHack does not store your agent chats and cannot access them.

FAQ: CVE-2026-96951 Questions Answered

What is CVE-2026-96951?

CVE-2026-96951 is a command injection (CWE-77) in Windows Shell that lets a crafted file evade Mark of the Web defenses. Microsoft rates it Important with a CVSS 3.1 score of 8.8, and its record says it is exploited. The user must open the file.

Is CVE-2026-96951 being exploited?

Yes. Microsoft’s record says exploited: Yes and “Exploitation Detected”. CISA added it to KEV on October 13, 2026, with an October 16 deadline and the forensic triage flag. No attribution or campaign details have been published.

What does a Mark of the Web bypass actually do?

It lets a file skip the warnings Windows attaches to downloads, such as SmartScreen checks, Protected View for Office documents and prompts for risky file types. It does not run the file for the attacker. A person still has to open it.

Which versions does CVE-2026-96951 affect?

Born City, citing Microsoft, says all supported Windows client and server versions are affected. NVD lists Windows 10 and 11 and Windows Server 2012 through 2025, and the fixed builds are in the table above, such as 10.0.26100.9711 for Windows 11 24H2.

Is CVE-2026-96951 linked to APT28?

Unknown. APT28 was tied to CVE-2026-21510 by Akamai, and CVE-2026-32202 was an incomplete patch of it, but no public source ties CVE-2026-96951 to APT28 or to any actor. Tenable’s Narang says it is unclear.

How do I check for CVE-2026-96951 exposure?

Confirm your Windows build is at or above the fixed build for your version, then review files that arrived without a Zone.Identifier tag from email or chat. No indicators of compromise have been published, so the file check is a heuristic.

The Bottom Line on CVE-2026-96951

CVE-2026-96951 removes a warning, not a person, which is why it scores 8.8 and still matters. Microsoft says it was exploited when the patch shipped, and it is the third exploited Windows Shell bypass of 2026.

Install the October updates, patch the machines that open outside files first, and do not treat Protected View as a control that this bug leaves intact. Then watch the earlier Shell bugs for a pattern: two of them were linked, and nobody has said yet whether this one makes three.


Categories

News

Previous

Pentest Report: What Auditors Want to See in It

Next

OSINT Tools for Bug Bounty: Match Each One to What a Program Will Pay For

On this page

What CVE-2026-96951 Actually Is

What Mark of the Web Does, and What CVE-2026-96951 Takes Away

How an Attack Using CVE-2026-96951 Would Reach a User

CVE-2026-96951 Is the Third Windows Shell Bug Exploited This Year

The first two are linked, the third is not yet

What the Public Record Does Not Tell You About CVE-2026-96951

Are You Exposed to CVE-2026-96951?

Fixing CVE-2026-96951

Hunting for CVE-2026-96951 Activity

The Pattern: A Name Search Would Miss These

Where XHack Fits After CVE-2026-96951

FAQ: CVE-2026-96951 Questions Answered

What is CVE-2026-96951?

Is CVE-2026-96951 being exploited?

What does a Mark of the Web bypass actually do?

Which versions does CVE-2026-96951 affect?

Is CVE-2026-96951 linked to APT28?

How do I check for CVE-2026-96951 exposure?

The Bottom Line on CVE-2026-96951

Related articles

Continue reading

CVE-2025-61624: The Medium-Rated FortiOS Bug Behind the PEERSTITCH Implant

News

CVE-2025-61624: The Medium-Rated FortiOS Bug Behind the PEERSTITCH Implant

CVE-2025-61624 is a medium-rated FortiOS path traversal that gave attackers root and planted PEERSTITCH. See which versi...

Read article
CVE-2026-110447: A Nextcloud Collabora Command Injection Fixed as a Tidy-Up Commit

News

CVE-2026-110447: A Nextcloud Collabora Command Injection Fixed as a Tidy-Up Commit

CVE-2026-110447 is an exploited Nextcloud Collabora command injection fixed as a tidy-up commit. See the timeline, the p...

Read article
PHP2Shell: Inside CVE-2026-109618 and CVE-2026-109619, the Core PHP RCE That Reaches Back to 2012

News

PHP2Shell: Inside CVE-2026-109618 and CVE-2026-109619, the Core PHP RCE That Reaches Back to 2012

PHP2Shell is an unauthenticated RCE chain in the PHP interpreter itself, confirmed against mod_php. Root cause, exploit ...

Read article