CVE-2023-0191

What it actually does

Out-of-bounds memory access in the kernel-mode handler of the NVIDIA GPU display driver allows local code to read or write beyond allocated buffers, resulting in denial of service or arbitrary data modification.

Why it matters to the business

Systems using affected NVIDIA GPU drivers may experience unexpected crashes or silent corruption of graphics-related data, leading to downtime on workstations, virtualized environments, and servers that rely on GPU acceleration.

How an attack unfolds

An attacker with local low-privileged access crafts input that triggers the vulnerable kernel handler, causing an out-of-bounds access that either crashes the driver or tampers with memory contents; no remote vector or public exploit code is known.

How to fix it

Install the latest NVIDIA GPU display driver for each affected platform (Windows, Linux, and virtual GPU products) as soon as vendor patches become available; apply hypervisor and guest OS updates that incorporate the fixed driver in virtualized deployments.

How to confirm the fix worked

Compare the installed NVIDIA driver version against the vendor's security bulletin for the affected product; re-test GPU-accelerated workloads after upgrade to confirm stability and absence of prior crash signatures.

Typical attack path

1. Obtain local interactive or service-level access on a host with an affected NVIDIA driver
2. Load or interact with the GPU display driver through user-mode APIs or direct device I/O
3. Supply malformed parameters that reach the kernel-mode out-of-bounds handler
4. Trigger denial of service via driver crash or achieve limited data tampering in kernel memory

How to detect it

Review system and driver logs for unexpected NVIDIA kernel-mode crashes or GPU reset events around the time of suspected activity; correlate with authentication logs showing local sessions on GPU-equipped hosts.

What your security team should do

• Prioritize driver updates on all systems listed in the affected products inventory
• Restrict local user access to GPU devices where operationally feasible
• Monitor vendor driver release channels for the corrected versions and schedule staged rollouts
• Include NVIDIA driver version checks in endpoint configuration compliance scans