CVE-2023-31018

What it actually does

Unprivileged local user triggers NULL-pointer dereference in the kernel-mode layer of the NVIDIA GPU driver, resulting in denial of service.

Why it matters to the business

Local users can crash systems with NVIDIA GPUs, causing unplanned downtime and service interruptions in physical, virtualized, and cloud environments where these drivers are deployed.

How an attack unfolds

An attacker with local unprivileged access on an affected system interacts with the NVIDIA GPU kernel driver to supply input that dereferences a null pointer, crashing the kernel or the affected process.

How to fix it

Apply the latest NVIDIA GPU driver updates for Windows and Linux on all affected products including virtual GPU, Ubuntu, Red Hat Enterprise Linux, and hypervisor platforms; reboot systems after installation and verify driver version.

How to confirm the fix worked

Confirm updated driver versions via vendor tools or package managers, review release notes for the fix, and test that normal GPU workloads no longer produce kernel crashes under the previous trigger conditions.

Typical attack path

1. Obtain local unprivileged user access on a system with vulnerable NVIDIA GPU driver
2. Locate and open the GPU driver device interface
3. Craft and submit a request that triggers the NULL dereference in kernel mode
4. Observe kernel panic or driver-induced denial of service

How to detect it

Review system and kernel logs for NVIDIA driver entries coinciding with unexpected reboots or crashes; correlate with user activity timelines on affected hosts.

What your security team should do

• Patch NVIDIA GPU drivers across all endpoints and servers as the primary remediation
• Limit local account creation and privilege escalation paths on GPU-equipped systems
• Monitor kernel logs for NVIDIA-related crashes as a temporary detection measure