Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain local low-priv access
-
DeliverySend crafted request to kernel module
-
ExploitTrigger integer underflow (CWE-191)
-
ExecutionCorrupt kernel memory
-
ImpactEscalate to kernel/SYSTEM privileges
Vulnerability Assessment
| Weakness |
CWE
ASVS 5.0
V15 - Secure Coding and Architecture Integer underflow checks
|
| Exploitation Conditions | Requires local, authenticated access to a Windows host running the affected NVIDIA GPU Display Driver (GeForce, RTX/Quadro/NVS, or Tesla) - the attacker must already be able to execute code as a low-privileged user (PR:L) to reach the vulnerable kernel module and supply the input that triggers the integer underflow. No user interaction and no special driver configuration is stated as required; the attack is not network-reachable (AV:L limits it to local access). Realistic risk is concentrated on shared, multi-user, VDI, and cloud-GPU deployments where an unprivileged tenant is present on the same host; single-user desktops where the attacker would already need a foothold face lower marginal risk. |
| Impact | The provided CVSS 3.1 vector (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H = 7.8 HIGH) is internally consistent with the description: this is a LOCAL, low-complexity, low-privilege kernel-mode flaw, NOT a remote one. The 'RCE' tag is therefore misleading - exploitation requires the attacker to already have local, authenticated code-execution on the host (PR:L), so this is a privilege-escalation / local-code-execution primitive rather than a network-reachable RCE. No EPSS score, no CISA KEV listing, and no public POC are present in the input, so there is no signal of active or imminent exploitation; this reads as a real-but-not-urgent local kernel bug patched proactively (NVIDIA-reported via product-security bulletin 5861). SSVC-style read: exploitation=none observed, automatable=no (local access prerequisite), technical impact=total (full C/I/A on the affected component). Priority is moderate: high technical impact but gated behind local access, which matters most on multi-user, VDI, or cloud-GPU hosts where an unprivileged tenant could escalate to kernel. The 'Microsoft' vendor tag is incorrect - the affected product is an NVIDIA driver that merely runs on Windows. |
| Exploit ScenarioDerived | A local, low-privileged user (or malware already running in a standard user context) on a Windows machine with an NVIDIA GPU driver issues crafted input to the vulnerable kernel module, triggering an integer underflow that corrupts kernel memory. By grooming the resulting out-of-bounds condition, the attacker escalates to SYSTEM/kernel privileges or crashes the driver for denial of service. No public exploit code is identified at time of analysis, and network access is not sufficient - local access is required. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 30, 2026
An integer underflow (CWE-191) in the kernel-mode component of the NVIDIA GPU Display Driver for Windows lets a local, authenticated low-privileged user corrupt kernel memory in a way that can lead to code execution, privilege escalation, denial of service, information disclosure, or data tampering (CVSS 3.1: 7.8, AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H). The issue affects the GeForce, RTX/Quadro/NVS and Tesla Windows driver branches and is not network-reachable, so an attacker must already be able to run code as a normal user on the host. There is no public exploit identified at time of analysis and no CISA KEV listing; the practical risk is concentrated on multi-user, VDI and cloud-GPU hosts where an unprivileged tenant shares the same machine as the vulnerable driver, while single-user desktops where the attacker must already hold a foothold face lower marginal risk.
RemediationAI
The primary fix is to update the NVIDIA Windows display driver to the patched release for the affected branch (GeForce Game Ready/Studio, NVIDIA RTX/Quadro Enterprise, or Data Center/Tesla), which NVIDIA documents in the advisory at https://github.com/NVIDIA/product-security/tree/main/2026/5861; because the supplied intelligence does not enumerate an exact fixed version number, verify the affected-to-fixed mapping in that bulletin and deploy the corresponding driver rather than assuming a specific build. Where immediate patching is not feasible, use host-level containment: on multi-user, VDI and cloud-GPU platforms, eliminate the low-privileged-tenant precondition by enforcing dedicated GPU passthrough or MIG partitioning, or by blocking untrusted local logon/interactive sessions on hosts with the vulnerable driver - accepting the trade-off of reduced consolidation density and higher infrastructure cost. Restricting local interactive access and removing unnecessary local accounts reduces the attack surface but does not remove the flaw. Because the vulnerable path requires local code execution (AV:L/PR:L), network-based mitigations such as perimeter filtering are ineffective and should not be relied on. Monitor for unexpected kernel crashes, bugchecks, or abnormal display-driver behavior as a possible indicator of attempted triggering, and note that no public exploit code was identified at time of analysis, so exploitation currently implies a targeted or capability-limited actor rather than commodity tooling.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 3 packages in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
The flaw lives in the kernel-mode portion of the NVIDIA Windows display driver stack (nvlddmkm.sys and the associated user-to-kernel IOCTL/D3DKMT escape surface that a local process can invoke without administrative rights). CWE-191 (Integer Underflow) describes the root-cause class: an arithmetic operation on a length, count, or offset field supplied through a driver interface wraps below zero instead of being rejected, so the driver subsequently performs a bounds check with a wrapped value, under-allocates a buffer, or computes an out-of-range index. The result is memory corruption in kernel space, which is why the assessed impact spans confidentiality, integrity and availability at High (C:H/I:H/A:H) with local, low-complexity, low-privilege prerequisites (AV:L/AC:L/PR:L/UI:N) and unchanged scope. The supplied CPE data covers cpe:2.3:a:nvidia:geforce, cpe:2.3:a:nvidia:rtx,_quadro,_nvs and cpe:2.3:a:nvidia:tesla, all with wildcarded versions, indicating that the advisory spans multiple consumer and enterprise driver branches rather than a single product build. Because the vulnerable code runs at ring 0, successfully triggering the underflow gives an attacker the ability to overwrite kernel structures, which is the mechanism behind the described code execution, escalation of privileges, denial of service, information disclosure and data tampering outcomes.
Know when Windows gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Local kernel-mode flaw reachable by a low-priv user without interaction (AV:L/AC:L/PR:L/UI:N); successful kernel memory corruption yields full C/I/A within the affected component (S:U).
Primary rating from Vendor (nvidia).
CVSS VectorVendor: nvidia
DescriptionCVE.org
NVIDIA GPU Display Driver for Windows contains a vulnerability in the kernel module where an attacker could cause an integer underflow. A successful exploit of this vulnerability might lead to code execution, denial of service, escalation of privileges, information disclosure, and data tampering.
Articles & Coverage 3
Executive ViewAI
NVIDIA GPU Display Driver for Windows contains a high-severity kernel memory corruption vulnerability that allows an authenticated local user with normal privileges to escalate to full control of the host. This poses significant risk to multi-user environments such as VDI and cloud-GPU platforms, where an unprivileged tenant could compromise other tenants' data or cause system-wide outages. With no vendor patch available, leadership should prioritize strict access controls and heightened monitoring until a fix is released.
No vendor patch available; restrict local interactive access to trusted users, enforce least privilege, implement enhanced monitoring for kernel-level anomalies, and isolate multi-tenant GPU hosts where feasible.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Integer overflow occurs when an arithmetic operation produces a result that exceeds the maximum value a given integer type can store.
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
Denial of Service attacks render applications or systems unavailable by overwhelming resources or triggering failure conditions.
Unauthenticated remote code execution in SmarterTools SmarterMail lets an attacker upload arbitrary files to any locatio
Windows Win32k fails to properly handle objects in memory, allowing local privilege escalation exploited in the wild in
A privilege escalation vulnerability (CVSS 5.5). Risk factors: actively exploited (KEV-listed), EPSS 94% exploitation pr
Windows Internet Shortcut Files (.url) contain an external control vulnerability (CVE-2025-33053, CVSS 8.8) that enables
Windows SMB contains an improper access control vulnerability (CVE-2025-33073, CVSS 8.8) enabling authenticated attacker
Twonky Server 8.5.2 on Linux and Windows allows unauthenticated access to the admin log file through a web service API b
An unrestricted file upload vulnerability exists in MiniWeb HTTP Server <= Build 300 that allows unauthenticated remote
FreeFloat FTP Server contains multiple critical design flaws that allow unauthenticated remote attackers to upload arbit
Serviio Media Server versions 1.4 through 1.8 on Windows contain an unauthenticated command injection in the /rest/actio
Twonky Server 8.5.2 uses hard-coded cryptographic keys for encrypting the administrator password. Combined with the cred
Mako Server versions 2.5 and 2.6 contain an unauthenticated OS command injection via the tutorial interface at examples/
React Native Metro Development Server binds to external interfaces by default and contains an OS command injection endpo
EUVD-2026-90128
GHSA-398p-6fwq-29q8
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB