NVIDIA Display Driver EUVD-2026-90097

Exploitation requires local access to a machine running an affected NVIDIA GPU Display Driver (GeForce, RTX/Quadro/NVS, or Tesla families) on Windows or Linux, plus the ability to run code at low/standard user privilege (PR:L) to reach the kernel-mode driver interface. No user interaction is needed (UI:N) and no special non-default configuration is called out - the vulnerable kernel mode layer is part of the standard driver install. What limits exploitation is the local-only attack vector (AV:L): the attacker must already have a foothold on the host; this cannot be triggered remotely. The 'Guest Driver' tag suggests virtualized/vGPU or guest-VM driver deployments are also in scope, which is relevant for GPU-accelerated VDI and datacenter environments.

This is a genuine, high-priority local privilege-escalation class flaw despite lacking sensational signals. The CVSS 3.1 base score of 7.8 (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) accurately reflects a locally-exploitable, low-complexity, low-privilege kernel-mode out-of-bounds write (CWE-787) with full high impact across confidentiality, integrity, and availability. The local attack vector (AV:L) and requirement for existing low-privilege access (PR:L) mean this is not a remote/wormable threat - it is a post-access escalation primitive, which is why the score is 7.8 rather than critical. 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. There is no SSVC data supplied. The signals are internally consistent (local kernel OOB write → PR:L, S:U, full CIA impact), so there is no CVSS-versus-description conflict to flag. NVIDIA GPU driver kernel flaws are historically attractive to LPE exploit developers and are frequently chained after initial access (e.g., from a sandboxed process or a foothold on a GPU-equipped workstation/server), so while raw exploitation probability data is absent, the practical risk to affected fleets is real and warrants prompt patching.

A local attacker who already has an unprivileged foothold on a Windows or Linux host with a vulnerable NVIDIA GPU driver issues crafted requests to the kernel-mode driver, triggering an out-of-bounds write in kernel memory. By controlling the overwritten kernel data, they corrupt kernel structures to escalate to SYSTEM/root, achieve kernel code execution, or crash the machine. No public exploit code was identified in the provided data, and exploitation requires prior local access and low-level privileges.