Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessGain low-privileged local/guest code execution
-
DeliverySend crafted IOCTLs to kernel-mode driver
-
ExploitTrigger use-after-free on freed kernel object
-
ExecutionReclaim/groom freed allocation
-
PersistHijack kernel control flow
-
ImpactEscalate to SYSTEM or crash host
Vulnerability Assessment
| Weakness |
CWE
ASVS 5.0
V15 - Secure Coding and Architecture Memory lifecycle management
MITRE ATT&CK
2 techniques across 2 tactics: Execution, Privilege Escalation The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires local code execution on a Windows host that has the affected NVIDIA GPU Display Driver (GeForce, RTX/Quadro/NVS, or Tesla branches) installed, plus a low-privileged account or sandboxed process able to open a handle to and issue calls against the kernel-mode driver interface. Per the CPE and tags, the vGPU/Guest Driver and Virtual GPU Manager components are also in scope, so a guest in a virtual-GPU deployment could target the kernel layer. Limiting factors: no remote vector (AV:L - must already be on the machine), and PR:L authentication is needed; there is no user interaction requirement (UI:N) and complexity is low (AC:L), so once local access exists the barrier is small. Hosts without an NVIDIA GPU/driver installed are unaffected. |
| Impact | This is a genuine, actionable local privilege-escalation risk rather than an inflated high-CVSS score. The CVSS 3.1 vector (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H = 7.8) is internally consistent with the CWE-416 use-after-free in the kernel-mode layer of the NVIDIA GPU Display Driver: a UAF in ring-0 code plausibly yields full C:H/I:H/A:H because arbitrary kernel memory corruption can escalate to SYSTEM. The local attack vector (AV:L) and required low privileges (PR:L) mean this is a post-compromise escalation primitive, not a remotely reachable flaw, which correctly moderates the score into the High (not Critical) band. No EPSS score, KEV listing, or public PoC was provided in the input, so exploitation status is 'no public exploit identified at time of analysis.' The absence of KEV/EPSS/PoC signals is consistent with a vendor-reported (NVIDIA product-security) memory-corruption bug patched proactively rather than one under active attack. Kernel UAFs are historically weaponizable by skilled attackers, so the practical risk is meaningful for multi-user or sandbox-escape scenarios even without confirmed exploitation. Note the description also lists 'code execution' and 'information disclosure' as possible impacts; the CVSS treats these as the local-scope C/I/A:H outcome (S:U), meaning impact stays within the compromised OS rather than crossing a security boundary to another system. |
| Exploit ScenarioDerived | A low-privileged local user (or code running in a sandbox/guest) on a Windows machine with the affected NVIDIA driver installed issues crafted IOCTLs/API calls to the kernel-mode driver to trigger a use-after-free, then grooms the freed kernel allocation to hijack control flow or corrupt kernel structures. Successful exploitation escalates to SYSTEM, enables kernel code execution, or crashes the host (DoS). No public exploit code was identified in the provided data, and no user interaction is required (UI:N). |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 30, 2026
Local privilege escalation in the NVIDIA GPU Display Driver for Windows allows a low-privileged local user - or a sandboxed process or guest with access to the driver interface - to trigger a use-after-free in the kernel-mode layer, potentially achieving kernel code execution, privilege escalation, denial of service, information disclosure, or data tampering. The assessed vector (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) means the attacker must already have code running on a host with an affected NVIDIA driver installed, but needs no user interaction and faces low complexity once local access exists; per the affected CPEs, the GeForce, RTX/Quadro/NVS, and Tesla branches are in scope, and the associated Guest Driver and Virtual GPU Manager tags mean a virtual-GPU guest may also be able to reach the kernel layer. No public exploit code or confirmed active exploitation was identified at the time of analysis, and hosts without an NVIDIA GPU/driver installed are unaffected.
RemediationAI
Patch status: the input data references NVIDIA's product-security bulletin (https://github.com/NVIDIA/product-security/tree/main/2026/5861) but does not specify an exact fixed driver version, so no patched version number is asserted here - consult that advisory and NVIDIA's driver download portal to identify and install the corrected Windows display driver for the applicable branch (GeForce, RTX/Quadro/NVS, or Tesla), then verify the installed driver version matches the advisory's fixed release and reboot to load the new kernel-mode driver. Because the flaw requires local access and code execution already running on the host, prioritize patching multi-user systems, VDI/vGPU hosts, and any machine where untrusted users or workloads share the GPU. Compensating controls while patching, with trade-offs: for vGPU deployments, stop assigning vGPU profiles or GPU passthrough to untrusted guests or move untrusted workloads to non-GPU VMs (side effect: loss of GPU acceleration for those guests, and existing profiles must be reconfigured); on shared workstations, restrict interactive or remote logon to fully trusted users and block sandboxed/untrusted applications from running on GPU-enabled endpoints (side effect: reduced usability for legitimate multi-user or developer scenarios); keep the driver updated to the latest branch release even if it precedes the advisory fix, as driver hardening often removes adjacent memory-safety bugs. There is no configuration switch that disables the kernel-mode driver while retaining GPU function, so a software-only workaround is not practical; detection should focus on unusual kernel driver interaction (repeated TDR/GPU crash events in the Windows event log) followed by unexpected SYSTEM-level process creation, which are the observable precursors and post-exploitation indicators for this class of local privilege-escalation bug.
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 is a use-after-free (CWE-416) in the kernel-mode layer of the NVIDIA Windows GPU display driver stack. In this architecture, user-mode graphics, CUDA, and management APIs communicate with a kernel-mode driver component through device handles and I/O control requests; the kernel layer owns GPU object lifetimes, memory allocations, and mapping bookkeeping. A use-after-free occurs when an object is released while a stale reference remains, and attacker-influenced operations or racing calls then dereference the freed memory - a class that in kernel context commonly yields arbitrary write/read primitives, pool corruption, or a reliable bugcheck (DoS). Because the defect lives in the kernel-mode interface rather than a user-mode library, any local principal that can open the driver's device interface and issue valid requests - including a compromised browser/process or a guest in a vGPU deployment, given the Guest Driver and Virtual GPU Manager tags and the CPE families listed (GeForce, RTX/Quadro/NVS, Tesla) - is a candidate trigger. All version ranges are wildcarded in the CPE data, so the exact vulnerable driver branches are not enumerated here and must be confirmed against NVIDIA's bulletin.
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
Kernel-mode UAF exploitable by an authenticated local low-priv user (AV:L/PR:L) with no interaction; kernel compromise yields full local C/I/A:H, scope unchanged.
Primary rating from Vendor (nvidia).
CVSS VectorVendor: nvidia
DescriptionCVE.org
NVIDIA GPU Display Driver for Windows contains a vulnerability in the kernel mode layer where an attacker could cause a use-after-free. A successful exploit of this vulnerability might lead to code execution, denial of service, or escalation of privileges, information disclosure, and data tampering.
Articles & Coverage 3
Executive ViewAI
A high-severity flaw in the NVIDIA GPU Display Driver for Windows lets any user or process already running on an affected machine escalate to kernel-level control of the operating system. Because the driver ships with GeForce, RTX/Quadro/NVS, and Tesla products - and can be reached from virtual-GPU guests - the exposure spans employee workstations, engineering and design fleets, VDI pools, and GPU-accelerated servers rather than a single niche system. Leadership should care because a successful attack converts limited local access into full host compromise, enabling data theft, tampering, or disruptive crashes, and NVIDIA has not yet released a fix, leaving the estate exposed until mitigation or patching is possible.
No vendor-released patch identified at time of analysis; interim mitigations: restrict which users, services, and sandboxed or guest processes can reach the NVIDIA driver interface; enforce least privilege and remove local admin rights from non-essential accounts; isolate or avoid running untrusted workloads on hosts with the affected driver; segment and monitor GPU-enabled servers and VDI hosts for anomalous kernel activity, crashes, or privilege escalation; track NVIDIA/OEM advisories for an exact fix version and apply immediately upon release.
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
Use-after-free vulnerabilities occur when a program continues to access memory through a pointer after that memory has been deallocated.
Memory corruption occurs when a program writes data beyond the boundaries of allocated memory regions or accesses memory in unintended ways, violating the integrity of the process's address space.
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-2026-90093
GHSA-9w9c-r99q-v2vj
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB