← Back to Feed CACHED · 2026-09-10 02:28:45 · CACHE_KEY CVE-2026-69845
CVE-2026-69845 · CWE-20 · Disclosed 2026-09-08

Heap-based buffer overflow in Windows DHCP Server

ASSESSED — NOISGATE V0.5
Vendor
Reassessed
Verdict:
Do you agree?
01 · The Real Story

It's like leaving the master key under the doormat — but only for people already inside the building

CVE-2026-69845 is a heap-based buffer overflow in the Windows DHCP Server service. An unauthenticated attacker on the same network segment can send specially crafted DHCP packets to trigger the overflow and achieve arbitrary code execution as SYSTEM. The flaw affects Windows Server editions running the DHCP Server role — Server 2016, 2019, 2022, and 2025 are all in scope. A companion vulnerability, CVE-2026-72979, targets the same attack surface with a use-after-free. Both were patched in the September 2026 Patch Tuesday cycle.

Microsoft stamped this CRITICAL / 9.8 with a CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H vector, but the AV:N designation is misleading. DHCP operates on UDP port 67, a broadcast/LAN protocol that is not routable across the internet by design. No legitimate enterprise exposes DHCP to the WAN. The real attack vector is *adjacent network* (AV:A), which would drop the CVSS to roughly 8.8. Microsoft itself rates exploitation as "Less Likely", there is no public PoC, no KEV listing, and no observed in-the-wild activity. The vendor severity overstates the reachability, but the *impact* — SYSTEM on infrastructure — is accurately severe.

"LAN-only attack surface caps real risk below vendor 9.8, but SYSTEM on infra is no joke."
02 · The Attack Path

4 steps from start to impact.

STEP 01

Gain internal network position

The attacker must first obtain a foothold on the same Layer 2 segment or VLAN as the target DHCP server. This could come from a prior compromise (phishing → workstation → lateral movement), a rogue device on a physical port, or a VPN with split-tunnel access to infrastructure VLANs. DHCP traffic is UDP broadcast; it does not traverse routers unless a DHCP relay agent is configured.
Conditions required:
  • Internal network access to a VLAN reachable by the DHCP server
  • Ability to send UDP packets on port 67
Where this breaks in practice:
  • Requires prior compromise or physical access — the attacker is already post-initial-access
  • Well-segmented networks isolate DHCP servers on management VLANs unreachable from user segments
  • 802.1X / NAC solutions prevent rogue devices from joining infrastructure VLANs
Detection/coverage: NAC logs, 802.1X authentication failures, switch port security alerts
STEP 02

Send malformed DHCP packets

From the internal position, the attacker crafts a DHCP DISCOVER or REQUEST message with oversized or malformed option fields designed to trigger the heap-based buffer overflow in dhcpssvc.dll. No authentication or credentials are required — the DHCP protocol has no auth mechanism by design. The packet must be precisely crafted to achieve controlled heap corruption rather than a simple crash.
Conditions required:
  • Knowledge of the heap layout and overflow trigger (no public PoC exists as of 2026-09-10)
  • Target server must be running the DHCP Server role and actively listening
Where this breaks in practice:
  • No public exploit or PoC code exists yet — attacker must develop the exploit from the patch diff
  • Heap-based overflows on modern Windows Server require ASLR/CFG bypass, raising the complexity bar
  • Windows Defender Exploit Guard and hardware-enforced stack protection add further friction
Detection/coverage: Snort 2 SIDs 67011-67032, 67036-67084; Snort 3 SIDs 301619-301629, 301632-301655; IDS/IPS monitoring DHCP traffic anomalies
STEP 03

Achieve SYSTEM-level code execution

Successful exploitation grants arbitrary code execution in the context of the DHCP Server service, which runs as NT AUTHORITY\SYSTEM. The attacker now has full control of the DHCP server host with the highest local privilege level.
Conditions required:
  • Successful heap corruption and control-flow hijack
  • Bypass of ASLR, CFG, and any exploit mitigations on the target
Where this breaks in practice:
  • Modern Windows exploit mitigations (ASLR, CFG, CET) significantly raise the bar for reliable exploitation
  • EDR agents on the DHCP server should flag anomalous process creation from dhcpssvc.dll context
Detection/coverage: EDR behavioral detection for anomalous child processes of svchost.exe hosting DHCP; Windows Event Log anomalies
STEP 04

Pivot and weaponize DHCP position

With SYSTEM on the DHCP server, the attacker can inject malicious DHCP options (rogue DNS, WPAD proxy, gateway) into lease responses to redirect traffic for every client on the scope. If the DHCP server is co-hosted on a domain controller — common in small-to-mid-size deployments — this is immediate domain compromise. Even on a standalone member server, the attacker has a trusted infrastructure pivot with broad L2 visibility.
Conditions required:
  • Maintained SYSTEM shell on the DHCP host
  • DHCP scopes actively serving clients
Where this breaks in practice:
  • DHCP Snooping on managed switches limits rogue DHCP response impact
  • DNSSEC and DoH reduce DNS hijack effectiveness
  • Separate DHCP from DC is a well-known best practice (though not universally followed)
Detection/coverage: DHCP audit logging, DHCP snooping alerts, DNS resolution anomaly detection, DC integrity monitoring
03 · Intelligence Metadata

The supporting signals.

In-the-wild exploitationNone observed. Not listed in CISA KEV. No GreyNoise activity. Microsoft rates exploitation as "Less Likely".
Proof-of-conceptNo public PoC as of 2026-09-10. No exploit code found on GitHub, ExploitDB, or researcher disclosures. Exploitation requires reverse-engineering the patch diff.
EPSS0.00954 (~95th percentile). Elevated relative to the general population but not in the top tier of imminent-exploitation indicators.
KEV statusNot listed. No CISA KEV entry as of 2026-09-10.
CVSS vectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H9.8. The AV:N is technically accurate per CVSS spec but practically misleading; DHCP is a LAN protocol. A more realistic vector would use AV:A, yielding ~8.8.
Affected versionsWindows Server 2016, 2019, 2022, and 2025 with the DHCP Server role installed and active. Client editions are not affected (they do not run the DHCP Server service).
Fixed versionsSeptember 2026 Patch Tuesday cumulative updates. Qualys QIDs: 110534, 110535, 110536, 388653, 388654, 388660, 388661, 50150, 92451-92464.
Internet exposureEffectively zero. DHCP operates on UDP/67, a broadcast protocol not routable over the internet. Shodan/Censys show negligible internet-facing DHCP servers. This is an internal-only attack surface.
Disclosure date2026-09-08 (September 2026 Patch Tuesday). No prior public disclosure.
Related CVEsCVE-2026-72979 — companion DHCP Server RCE (use-after-free), also CVSS 9.8, same patch cycle. Both should be remediated together.
04 · The Call

noisgate verdict.

Final Verdict
DOWNGRADED to HIGH (8.5/10)

The single most decisive factor is that DHCP is a LAN-only protocol with effectively zero internet exposure, meaning the CVSS AV:N vector overstates reachability — any attacker exploiting this is already post-initial-access on the internal network. The impact on infrastructure running as SYSTEM with potential DC co-hosting prevents further downgrade below HIGH.

HIGH Vulnerability severity and attack vector characterization
HIGH Absence of public PoC and in-the-wild exploitation
MEDIUM Affected version range (inferred from DHCP Server role availability across Server editions)

Why this verdict

  • LAN-only protocol caps reachability. DHCP operates on UDP/67, a broadcast protocol that does not traverse routers or the internet. The CVSS AV:N is technically correct but practically misleading — the real attack vector is adjacent network (AV:A). This single factor removes the entire remote-unauthenticated-over-internet threat model that justifies a 9.8.
  • No public exploit lowers immediacy. There is no PoC, no exploit code, and Microsoft rates exploitation as "Less Likely." Weaponizing a heap overflow on modern Windows Server with ASLR, CFG, and CET requires significant attacker investment. This is not a spray-and-pray vulnerability.
  • Role multiplier: DHCP server is infrastructure, sometimes co-hosted on DCs. (a) *Low-value role:* N/A — DHCP Server is not installed on workstations. (b) *Typical role:* Dedicated member server running DHCP — compromise yields SYSTEM on one infrastructure host with ability to poison DHCP options for the subnet. Blast radius: subnet-scale MITM. (c) *High-value role:* DHCP co-hosted on a domain controller (common in SME/branch offices, estimated 10-20% of AD deployments) — SYSTEM on DC = immediate domain compromise, fleet-scale blast radius. This high-value-role outcome floors the verdict at HIGH.

Why not higher?

Upgrading to CRITICAL would require either active exploitation, a public PoC lowering the exploitation bar, or internet-facing exposure. None of these conditions exist. The attack requires prior internal network access (post-initial-access), the exploit complexity for a heap overflow on modern Windows is non-trivial, and Microsoft's own assessment is "Exploitation Less Likely." The floor analysis keeps this at HIGH but the friction prevents it from reaching CRITICAL.

Why not lower?

Dropping to MEDIUM would ignore the blast radius. DHCP servers run as SYSTEM on infrastructure hosts that configure every endpoint's network stack. The co-hosting-on-DC scenario (estimated 10-20% of deployments) means successful exploitation can end in domain takeover. The vulnerability is unauthenticated once on-network, requires zero user interaction, and the DHCP protocol has no authentication mechanism — there is no credential barrier. The role-multiplier floor keeps this firmly at HIGH.

05 · Compensating Control

What to do — in priority order.

  1. Segment DHCP servers onto dedicated management VLANs — Restrict which hosts can reach UDP/67 on the DHCP server to only DHCP relay agents and authorized admin workstations. This eliminates the attack surface from compromised user endpoints. Deploy within 30 days per noisgate mitigation SLA for HIGH.
  2. Deploy IDS/IPS signatures for malformed DHCP traffic — Enable Snort SIDs 67011-67032, 67036-67084 (Snort 2) or 301619-301655 (Snort 3) to detect and block exploit attempts targeting the DHCP service. These should be deployed on sensors monitoring infrastructure VLANs.
  3. Separate DHCP Server role from Domain Controllers — If DHCP is currently co-hosted on DCs, migrate the role to dedicated member servers. This reduces the blast radius from domain-takeover to single-host compromise. This is a best-practice architectural change that should be prioritized for any environment where it hasn't been done.
  4. Enable DHCP Snooping on managed switches — DHCP Snooping validates DHCP messages and rate-limits DHCP traffic on untrusted ports. While it primarily defends against rogue DHCP servers, it also limits an attacker's ability to interact with the DHCP service from arbitrary switch ports.
  5. Ensure EDR coverage on DHCP server hosts — EDR agents should be active on all infrastructure servers including DHCP. Behavioral detection for anomalous process creation from svchost.exe hosting the DHCP service would catch post-exploitation activity.
What doesn't work
  • Network-level firewalling of UDP/67 from all hosts — this breaks DHCP for legitimate clients. You can restrict source VLANs but cannot block the port entirely without breaking IP address assignment.
  • Web Application Firewalls (WAFs) — DHCP is a UDP broadcast protocol, not HTTP. WAFs have zero visibility into this traffic.
  • MFA / credential hardening — the DHCP protocol has no authentication mechanism. The vulnerability is exploitable without any credentials, so identity-layer controls do not apply.
06 · Verification

Crowdsourced verification payload.

Run this on each Windows Server that might have the DHCP Server role installed. Execute in an elevated PowerShell session (Run as Administrator). No external dependencies required. Example: powershell -ExecutionPolicy Bypass -File .\Check-CVE-2026-69845.ps1

noisgate-verify.ps1
POWERSHELLREAD-ONLYSAFE
#Requires -RunAsAdministrator
<#
  Check-CVE-2026-69845.ps1
  Checks whether the Windows DHCP Server role is installed and whether
  the September 2026 cumulative update has been applied.
#>

$VulnCVE = 'CVE-2026-69845'

# Step 1: Check if DHCP Server role is installed
$dhcpFeature = Get-WindowsFeature -Name DHCP -ErrorAction SilentlyContinue
if (-not $dhcpFeature -or $dhcpFeature.InstallState -ne 'Installed') {
    Write-Host "PATCHED — DHCP Server role is not installed. $VulnCVE does not apply to this host."
    exit 0
}

Write-Host "[!] DHCP Server role is INSTALLED on this host."

# Step 2: Check if DHCP Server service is running
$dhcpSvc = Get-Service -Name DHCPServer -ErrorAction SilentlyContinue
$svcRunning = $dhcpSvc -and $dhcpSvc.Status -eq 'Running'

# Step 3: Check for September 2026 patches
# Look for cumulative updates dated 2026-09 or later
$patchDate = [DateTime]'2026-09-08'
$recentPatches = Get-HotFix | Where-Object {
    $_.InstalledOn -ge $patchDate
} | Sort-Object InstalledOn -Descending

if ($recentPatches.Count -gt 0) {
    $latestKB = $recentPatches[0].HotFixID
    $latestDate = $recentPatches[0].InstalledOn.ToString('yyyy-MM-dd')
    Write-Host "PATCHED — DHCP Server role is installed but September 2026 update ($latestKB, installed $latestDate) is present."
    exit 0
} else {
    $lastPatch = Get-HotFix | Sort-Object InstalledOn -Descending | Select-Object -First 1
    $lastDate = if ($lastPatch) { $lastPatch.InstalledOn.ToString('yyyy-MM-dd') } else { 'UNKNOWN' }
    $runningSuffix = if ($svcRunning) { ' and service is ACTIVELY RUNNING' } else { ' (service is stopped but role is installed)' }
    Write-Host "VULNERABLE — DHCP Server role is installed$runningSuffix. Last patch installed: $lastDate. September 2026 cumulative update NOT found."
    exit 1
}
07 · Bottom Line

If you remember one thing.

TL;DR
Patch your DHCP servers in the next regular maintenance window, but do not treat this as a drop-everything emergency. At noisgate's reassessed HIGH severity, your noisgate mitigation SLA is 30 days to deploy compensating controls — prioritize VLAN segmentation of DHCP servers and IDS signature deployment. Your noisgate remediation SLA is 180 days to apply the September 2026 cumulative update across all Windows Server hosts running the DHCP role. If any of your DHCP servers are co-hosted on domain controllers, treat those specific hosts as urgent and patch within the first maintenance window — the blast radius on a DC is domain takeover. Also patch companion CVE-2026-72979 in the same cycle. There is no active exploitation and no public PoC today, so you have time to test properly, but don't let this one age out.

Sources

  1. CrowdStrike — September 2026 Patch Tuesday Analysis
  2. Cisco Talos — September 2026 Patch Tuesday Snort Rules
  3. Zero Day Initiative — September 2026 Security Update Review
  4. Rapid7 — September 2026 Patch Tuesday
  5. Qualys — September 2026 Patch Tuesday Security Update Review
  6. Microsoft MSRC — CVE-2026-69845 Update Guide
  7. Action1 — September 2026 CISO Executive Summary
Peer Review

What defenders are saying.

Submit a review attribution: handle + country only
0 flags selected · stored anonymously
Validation Results

Crowdsourced verification outputs.

Results submitted by users who ran the verification payload against their environment.