Six zero-day vulnerabilities. All actively exploited. One already weaponized since December 2025. And you have until March 3rd to patch them all.
If you manage Windows systemsβwhether a home PC, corporate endpoint, or enterprise server farmβstop what youβre doing and read this. Microsoftβs February 2026 Patch Tuesday, released on February 10th, represents one of the most dangerous security updates in recent memory. With six zero-day vulnerabilities under active exploitation in the wild, this isnβt a routine βpatch when convenientβ situation. This is a βpatch now or become a statisticβ emergency.
CISA has already added all six vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog with a hard deadline of March 3, 2026 for federal agencies. But make no mistakeβprivate organizations face the same threat actors using the same exploits. The only difference is you donβt have a federal mandate forcing compliance.
Letβs break down what youβre up against, which systems are affected, and exactly what to patch first.
The Headline Numbers
Before we dive deep, hereβs what February 2026 looks like by the numbers:
Metric February 2026 Context
Total CVEs Patched 54-59 Below average (typical: 80-120)
Zero-Days (Actively Exploited) 6 Tied for highest ever with March 2025
Publicly Disclosed Before Patch 3 Above average
Critical Severity 2-5 Below average
CISA KEV Additions 6 All zero-days added same day
The relatively low total CVE count is deceptive. What matters is the quality of vulnerabilities attackers are exploiting, not quantity. Six zero-days means threat actors had six different entry points into your systems before patches even existed.
For context: Microsoft addressed 41 zero-days across all of 2025. February 2026 alone accounts for roughly 15% of that totalβin a single month.
As Trend Microβs Zero Day Initiative bluntly put it: βThe number of bugs under active attack is extraordinarily high.β
Why This Patch Tuesday is Different
Every month brings security patches. What makes February 2026 exceptional?
1. Three Security Feature Bypasses That Make Phishing Actually Dangerous
Three of the six zero-days (CVE-2026-21510, CVE-2026-21513, CVE-2026-21514) bypass Windows security warningsβthe βAre you sure you want to run this?β dialogs that stand between your users and catastrophe. When these protections are bypassed, phishing attacks become nearly frictionless. Users click, malware runs, no warning appears.
2. Professional-Grade Exploits Already Circulating
When security researchers at 0patch (ACROS Security) discovered one of these vulnerabilities in December 2025, they found the exploit in a public malware repository. Their assessment? The code quality suggested βprofessional workββnot amateur script kiddie stuff, but sophisticated tooling likely developed by well-resourced threat actors.
3. Confirmed Six-Week Pre-Patch Exploitation
CrowdStrikeβs Advanced Research Team revealed something alarming: one of these zero-days (CVE-2026-21533) was actively used against U.S. and Canada-based entities since at least December 24, 2025. Thatβs more than six weeks of exploitation before patches became available. If you havenβt patched yet, you need to assume compromise and hunt for indicators.
4. Coordinated Discovery Suggests Coordinated Attacks
Multiple vulnerabilities were discovered by overlapping teamsβMicrosoft Threat Intelligence Center (MSTIC), Google Threat Intelligence Group (GTIG), Microsoft Security Response Center (MSRC), and CrowdStrike. When multiple premier threat intelligence teams simultaneously discover related vulnerabilities, it often indicates theyβre tracking the same advanced threat campaign.
The Six Zero-Days: Deep Dive
Letβs examine each vulnerability in detail, starting with the most urgent.
CVE-2026-21533: Windows Remote Desktop Services Elevation of Privilege
The one thatβs been exploited since December.
Attribute Details
CVSS Score 7.8 - 8.8 (High)
Severity Important
Affected Component Windows Remote Desktop Services (RDS)
Attack Vector Local
Publicly Disclosed No
Actively Exploited Yes
Exploitation Confirmed Since December 24, 2025
What It Does
This vulnerability allows an attacker with basic local access to escalate their privileges to SYSTEMβthe highest privilege level in Windows. The flaw exists in how Windows Remote Desktop Services manages privileges, specifically through improper handling of service configuration keys.
How Attackers Use It
According to CrowdStrikeβs technical analysis:
βThe CVE-2026-21533 exploit binary modifies a service configuration key, replacing it with an attacker-controlled key, which could enable adversaries to escalate privileges to add a new user to the Administrator group.β
Translation: Get basic access to a system (phishing, stolen credentials, whatever), run this exploit, become Administrator instantly.
Why Itβs Critical
CrowdStrike issued a stark warning about what happens now that the patch is public:
βMicrosoftβs public disclosure of CVE-2026-21533 will almost certainly encourage threat actors possessing CVE-2026-21533 exploit binaries, as well as any exploit brokers possessing the underlying exploit, to use or monetize the exploits in the near term.β
In other words: every criminal who had this exploit is now racing to use it before systems get patched. Every exploit broker is selling it at a premium. The clock is ticking.
Whoβs Affected
- All Windows versions with Remote Desktop Services- Windows Server (all supported versions)- Windows 10 and 11 workstations
Detection Note
Since this was exploited for 6+ weeks before patches, consider running threat hunting queries looking for:
- Unexpected Administrator group changes- Modified service configurations- New local accounts created in December 2025 - February 2026
CVE-2026-21510: Windows Shell Security Feature Bypass
The SmartScreen killer.
Attribute Details
CVSS Score 7.5 - 8.8 (High)
Severity Important
Affected Component Windows Shell
Attack Vector Network (requires user interaction)
Publicly Disclosed Yes
Actively Exploited Yes
What It Does
This vulnerability bypasses two critical Windows security features: SmartScreen and Windows Shell security prompts. These are the warnings that appear when you try to open a downloaded file or executable from the internetβthe Mark of the Web (MoTW) protection system.
When bypassed, downloaded malware can execute silently without any warning dialog.
How Attackers Use It
The attack flow is disturbingly simple:
- Attacker sends email with malicious
.lnk(shortcut) file2. User clicks the file3. Normally: Windows displays βThis file came from the internet and may be dangerousβ warning4. With this exploit: No warning appears5. Malicious payload executes immediately
Why Itβs Critical
SmartScreen and MoTW warnings are often the last line of defense against phishing. Theyβre the final speed bump between a userβs click and malware execution. Bypassing them makes social engineering attacks dramatically more effective.
Who Discovered It
- Microsoft Threat Intelligence Center (MSTIC)- Microsoft Security Response Center (MSRC)- Office Product Group Security Team- Google Threat Intelligence Group (GTIG)- Anonymous researcher
The multi-team discovery suggests this was found while tracking an active threat campaign.
CVE-2026-21513: MSHTML Framework Security Feature Bypass
The legacy browser component that wonβt die.
Attribute Details
CVSS Score 8.8 (High)
Severity Important
Affected Component MSHTML Framework (Internet Explorer Trident engine)
Attack Vector Network (requires user interaction)
Publicly Disclosed Yes
Actively Exploited Yes
What It Does
βWait, Internet Explorer is deadβwhy am I patching it?β
Great question. While Internet Explorer as a standalone browser is discontinued, its rendering engine (MSHTML/Trident) lives on throughout Windows. Itβs embedded in various applications for displaying HTML content, and itβs still present in every Windows installation.
This vulnerability bypasses security checks when MSHTML processes malicious content, potentially weakening browser or Office sandbox protections.
How Attackers Use It
According to Action1βs analysis:
βA crafted file can silently bypass Windows security prompts and trigger dangerous actions with a single click.β
Attack vectors include:
- Malicious HTML files- Crafted
.lnkshortcut files that leverage MSHTML for rendering- Any application using MSHTML to display web content
Why It Matters Even Though IE Is Dead
Every Windows system still has MSHTML components. Many enterprise applications still use MSHTML for embedded web rendering. Legacy doesnβt mean obsolete when it comes to attack surface.
CVE-2026-21514: Microsoft Word Security Feature Bypass
The Office document that disarms your defenses.
Attribute Details
CVSS Score 5.5 - 8.1 (varies by configuration)
Severity Important
Affected Component Microsoft Word, Microsoft 365, Microsoft Office
Attack Vector Local (requires user to open malicious file)
Publicly Disclosed Yes
Actively Exploited Yes
What It Does
This vulnerability bypasses OLE (Object Linking and Embedding) mitigationsβsecurity controls designed to protect users from malicious embedded objects in Office documents. It works by exploiting how Word makes security decisions based on untrusted input.
How Attackers Use It
Classic phishing attack flow:
- Attacker sends email with malicious Word document2. User opens document3. Embedded malicious object executes4. OLE mitigations that should have blocked execution are bypassed
Note: The Preview Pane is NOT an attack vector here. Users must actually open the document.
Historical Context
This vulnerability bears strong similarity to CVE-2026-21509, which Microsoft addressed with an emergency out-of-band update in January 2026 due to active exploitation. The pattern suggests attackers are systematically probing Officeβs OLE handling for bypass techniques.
CVE-2026-21519: Desktop Window Manager Elevation of Privilege
Type confusion leads to SYSTEM.
Attribute Details
CVSS Score 7.8 (High)
Severity Important
Affected Component Desktop Window Manager (DWM)
Attack Vector Local
Publicly Disclosed No
Actively Exploited Yes
What It Does
The Desktop Window Manager is a Windows service responsible for rendering your graphical user interfaceβwindows, transparency effects, animations. It runs on every Windows system with a GUI.
This vulnerability is a type confusion flaw where DWM treats one type of data as another, leading to unintended behavior that attackers can exploit for privilege escalation.
How Attackers Use It
- Attacker already has local access (low-privilege user)- Runs crafted exploit program- DWM processes malformed data- Type confusion triggers- Attacker gains SYSTEM privileges
No user interaction required once attacker has initial access.
Post-Exploitation Impact
With SYSTEM access, attackers can:
- Disable security tools (EDR, antivirus)- Deploy additional malware- Access credential stores- Create persistence mechanisms- Move laterally to other systems- Potentially compromise the entire domain
CVE-2026-21525: Windows Remote Access Connection Manager DoS
The VPN killer.
Attribute Details
CVSS Score 6.2 (Medium)
Severity Moderate
Affected Component Windows Remote Access Connection Manager (RasMan)
Attack Vector Local
Publicly Disclosed No
Actively Exploited Yes
What It Does
This is a null pointer dereference vulnerability in the RasMan serviceβthe Windows component that manages VPN and dial-up connections. A local attacker can crash the service, causing denial of service.
βA DoS vulnerability? Thatβs just an annoyance.β
Not in this case. Let me explain why this matters more than the βMediumβ severity suggests.
The VPN Cascade Problem
Automoxβs Ryan Braunstein highlighted a critical enterprise impact:
βOrganizations using always-on VPN face severe risk. VPN service crash causes endpoints with βfail closeβ policies to lose network access. IT teams cannot reach affected machines to patch them.β
The attack scenario:
- Attacker compromises endpoint (phishing, browser exploit, etc.)2. Runs simple script that crashes RasMan3. VPN connection dies4. Endpoint with βfail closeβ policy loses all network connectivity5. IT team canβt remotely access the machine to fix it6. Machine requires physical intervention
Scale this to hundreds of remote workers, and youβve got a logistical nightmare.
The Distraction Attack
Worse: crashing VPN services can serve as a distraction. While IT scrambles to restore connectivity for hundreds of disconnected users, attackers conduct actual exfiltration or lateral movement elsewhere in the network.
Discovery History
0patch (ACROS Security) discovered this exploit in December 2025 in a public malware repository while investigating a related vulnerability (CVE-2025-59230). The combined exploit quality suggested professional development.
Attack Chains: How These Vulnerabilities Work Together
Individual vulnerabilities tell one story. Attack chains tell the real story.
Chain 1: Phishing β Code Execution β Full Compromise
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β 1. INITIAL ACCESS β
β β’ Attacker sends email with malicious .lnk file β
β β’ CVE-2026-21510 bypasses SmartScreen warnings β
β β’ User clicks without seeing security prompts β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 2. EXECUTION β
β β’ Malware runs with user-level privileges β
β β’ Establishes initial foothold β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 3. PRIVILEGE ESCALATION β
β β’ CVE-2026-21519 (DWM) OR CVE-2026-21533 (RDS) β
β β’ Attacker escalates to SYSTEM β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 4. FULL COMPROMISE β
β β’ Disable security tools β
β β’ Dump credentials β
β β’ Deploy ransomware/backdoors β
β β’ Move laterally β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Chain 2: Document-Based Attack
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β 1. DELIVERY β
β β’ Malicious Word document via email/download β
β β’ "Invoice.docx" or "Report.docx" β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 2. BYPASS β
β β’ CVE-2026-21514 bypasses OLE mitigations β
β β’ Embedded malicious content executes β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 3. ESCALATION β
β β’ CVE-2026-21533 adds attacker to Admin group β
β β’ SYSTEM access achieved β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β 4. PERSISTENCE β
β β’ Create backdoor accounts β
β β’ Install remote access tools β
β β’ Establish command & control β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Chain 3: VPN Disruption + Exfiltration
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β PARALLEL ATTACK β
β β
β Thread A: DISTRACTION Thread B: REAL ATTACK β
β ββ Initial foothold ββ Already in network β
β ββ CVE-2026-21525 crash RasMan ββ Begin data exfiltrationβ
β ββ VPN dies enterprise-wide ββ Deploy ransomware β
β ββ IT scrambles to fix VPN ββ Destroy backups β
β ββ (They're not watching B) ββ Ransom note appears β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
CISA Deadline: March 3, 2026
On February 10, 2026βthe same day as Patch TuesdayβCISA added all six zero-days to its Known Exploited Vulnerabilities (KEV) catalog.
KEV Catalog Entries
CVE CISA Description Deadline
CVE-2026-21510 Microsoft Windows Shell Protection Mechanism Failure March 3, 2026
CVE-2026-21513 Microsoft MSHTML Framework Security Feature Bypass March 3, 2026
CVE-2026-21514 Microsoft Office Word Reliance on Untrusted Inputs March 3, 2026
CVE-2026-21519 Microsoft Windows Type Confusion Vulnerability March 3, 2026
CVE-2026-21525 Microsoft Windows NULL Pointer Dereference March 3, 2026
CVE-2026-21533 Windows Remote Desktop Services Elevation of Privilege March 3, 2026
Who Must Comply
Under Binding Operational Directive (BOD) 22-01, Federal Civilian Executive Branch (FCEB) agencies are required to:
- Apply patches by the specified due date2. Protect networks against attacks exploiting cataloged vulnerabilities3. Report compliance status
Private Sector Recommendation
CISA doesnβt just mandate federal complianceβthey explicitly recommend all organizations treat KEV entries seriously:
βCISA strongly urges all organizations to reduce their exposure to cyberattacks by prioritizing timely remediation of Catalog vulnerabilities as part of their vulnerability management practice.β
Translation: Just because youβre not legally required to patch by March 3rd doesnβt mean you shouldnβt. The same threat actors targeting federal agencies are targeting you.
Patch Prioritization: What to Fix First
Not all patches are created equal. Hereβs your prioritization guide based on exploitation risk, impact, and attack surface.
Tier 1: IMMEDIATE β Patch Within 24-48 Hours
These are actively exploited with public disclosure and/or confirmed widespread attacks.
CVE Component Why Immediate
CVE-2026-21533 Remote Desktop Services Exploited since December 2025; post-patch surge expected
CVE-2026-21510 Windows Shell Publicly disclosed; bypasses SmartScreen; phishing enabler
CVE-2026-21513 MSHTML Framework Publicly disclosed; bypasses security prompts
CVE-2026-21514 Microsoft Word Publicly disclosed; OLE bypass; document-based attacks
Action Steps:
# Check for pending updates
Get-WindowsUpdate -MicrosoftUpdate
# Install February 2026 cumulative update
Install-WindowsUpdate -AcceptAll -AutoReboot
# Verify installation
Get-HotFix | Where-Object {$_.InstalledOn -gt "2026-02-10"}
For enterprises using WSUS/SCCM/Intune:
# WSUS: Approve February 2026 updates for immediate deployment
# Classification: Security Updates
# Product: Windows 10, Windows 11, Windows Server, Microsoft Office
# Verify deployment status
Get-WsusUpdate | Where-Object {$_.UpdatesNeeded -gt 0}
Tier 2: HIGH PRIORITY β Patch Within 72 Hours
Actively exploited but requires local access. Part of attack chains.
CVE Component Why High Priority
CVE-2026-21519 Desktop Window Manager Grants SYSTEM; no user interaction needed once local
Why not Tier 1? Requires attacker to already have local access. However, combined with Tier 1 security bypasses, this completes compromise chains. Patch immediately after Tier 1.
Tier 3: PRIORITY β Patch Within 1 Week
Lower severity but operational impact. Actively exploited.
CVE Component Why Priority
CVE-2026-21525 Remote Access Connection Manager VPN disruption; distraction attacks; cascade failures
Why Tier 3? Medium severity, DoS only. However, if your organization relies on always-on VPN, move this to Tier 2.
Bonus: Critical Non-Zero-Day Patches
While focused on zero-days, donβt ignore these CVSS 9.8 critical vulnerabilities:
CVE Component CVSS
CVE-2026-21531 Azure SDK for Python 9.8 (Critical)
CVE-2026-24300 Azure Front Door 9.8 (Critical)
If you use Azure SDK for Python or Azure Front Door, patch these simultaneously with Tier 1.
Step-by-Step Remediation Guide
For Individual Users / Small Businesses
- Check for updates immediately:
- Settings β Windows Update β Check for updates- Install all available updates- Restart when prompted2. Update Microsoft Office:
- Open any Office app β File β Account β Update Options β Update Now- Or via Microsoft 365 admin center3. Verify patches applied:
- Settings β Windows Update β Update history- Look for β2026-02 Cumulative Updateβ
For Enterprise IT Teams
Phase 1: Emergency Deployment (Day 1-2)
# 1. Identify unpatched systems
# Example using SCCM/ConfigMgr
Get-CMDevice -CollectionName "All Windows Workstations" |
Where-Object {$_.LastUpdateScan -lt "2026-02-10"} |
Export-CSV "unpatched-systems.csv"
# 2. Deploy to high-risk systems first
# - Executive workstations
# - Systems with RDP enabled
# - User-facing endpoints (phishing targets)
# 3. Verify deployment
$unpatched = Get-ADComputer -Filter * | ForEach-Object {
$hotfix = Get-HotFix -ComputerName $_.Name -ErrorAction SilentlyContinue |
Where-Object {$_.InstalledOn -gt "2026-02-10"}
if (-not $hotfix) { $_.Name }
}
$unpatched | Out-File "still-unpatched.txt"
Phase 2: Validation (Day 2-3)
# Verify specific KB installation
$KB = "KB5034123" # Replace with actual KB number
Get-ADComputer -Filter * | ForEach-Object {
$result = Get-HotFix -ComputerName $_.Name -Id $KB -ErrorAction SilentlyContinue
[PSCustomObject]@{
Computer = $_.Name
Patched = [bool]$result
InstalledOn = $result.InstalledOn
}
} | Export-CSV "patch-status.csv"
Phase 3: Threat Hunting (Day 3-7)
Given CVE-2026-21533 was exploited since December 2025, hunt for compromise indicators:
# Check for recently added local administrators
$startDate = Get-Date "2025-12-01"
Get-WinEvent -FilterHashtable @{
LogName = 'Security'
ID = 4732 # Member added to security-enabled local group
} | Where-Object {
$_.TimeCreated -gt $startDate -and
$_.Message -match "Administrators"
} | Select-Object TimeCreated, Message
# Check for suspicious service modifications
Get-WinEvent -FilterHashtable @{
LogName = 'System'
ID = 7045 # Service installed
} | Where-Object {$_.TimeCreated -gt $startDate}
# Review new local accounts
Get-LocalUser | Where-Object {
$_.Enabled -eq $true -and
$_.LastLogon -gt $startDate
} | Select-Object Name, Enabled, LastLogon, PasswordLastSet
Beyond Patching: Additional Hardening
Immediate Mitigations (If Patching Is Delayed)
For CVE-2026-21533 (RDS):
# Disable Remote Desktop if not needed
Set-ItemProperty -Path 'HKLM:\System\CurrentControlSet\Control\Terminal Server' `
-Name "fDenyTSConnections" -Value 1
# Or restrict RDP to specific IPs via firewall
New-NetFirewallRule -DisplayName "RDP Restricted" `
-Direction Inbound -Protocol TCP -LocalPort 3389 `
-RemoteAddress "10.0.0.0/8" -Action Allow
For CVE-2026-21510/CVE-2026-21513 (Phishing bypasses):
- Enable Attack Surface Reduction (ASR) rules- Block .lnk files in email attachments- Increase email filtering sensitivity
# Enable ASR rules via Group Policy or PowerShell
Set-MpPreference -AttackSurfaceReductionRules_Ids `
BE9BA2D9-53EA-4CDC-84E5-9B1EEEE46550 `
-AttackSurfaceReductionRules_Actions Enabled
For CVE-2026-21514 (Word):
- Block macros from internet-sourced documents- Enable Protected View for Office files
Long-Term Hardening
- Implement least privilege: Users shouldnβt be local administrators2. Network segmentation: Limit RDP access to jump servers only3. EDR deployment: Ensure behavioral detection is active4. Security awareness: Retrain users on phishing given bypass vulnerabilities5. Patch automation: Donβt rely on manual patching for critical updates
Whatβs Coming Next
Secure Boot Certificate Expiration (June 2026)
Separate from this Patch Tuesday but important: Microsoft has begun rolling out updated Secure Boot certificates. The original 2011 certificates expire in late June 2026. Plan for this transition.
Exploit Surge Expected
CrowdStrikeβs warning bears repeating: Now that CVE-2026-21533 is public, expect a surge in exploitation attempts. Threat actors who had private access to this exploit are racing to monetize it before patch adoption increases.
March 2026 Patch Tuesday
Given Februaryβs zero-day count, expect heightened attention on Marchβs release. Mark your calendar for March 10, 2026.
Key Takeaways
- Patch immediately: Six zero-days are actively exploited. This is not a drill.2. CISA deadline is March 3, 2026: Federal mandate, but everyone should treat it as their deadline.3. Prioritize the security bypasses first: CVE-2026-21510, CVE-2026-21513, CVE-2026-21514 make phishing attacks silent. Patch these today.4. CVE-2026-21533 has been exploited since December 2025: If you havenβt patched, assume compromise and hunt for indicators.5. Attack chains are real: These vulnerabilities work together. A phishing email + security bypass + privilege escalation = full compromise.6. VPN users beware: CVE-2026-21525 can crash your VPN service. If you use always-on VPN, prioritize this patch.7. Donβt forget Azure: Two CVSS 9.8 critical vulnerabilities in Azure services need attention too.
Further Reading
- Microsoft Security Response Center - February 2026 Release Notes- CISA Known Exploited Vulnerabilities Catalog- CrowdStrike Patch Tuesday Analysis February 2026- Zero Day Initiative February 2026 Analysis
The bottom line: February 2026 Patch Tuesday isnβt business as usual. Six zero-days, all actively exploited, all added to CISAβs KEV catalog on day one. One has been exploited for over six weeks against U.S. and Canadian organizations.
Stop reading. Start patching.
Stay safe out there.