NVIDIA GPU Driver CVE-2026-47529

Requires local access to a Linux host running an affected NVIDIA GPU Display Driver (GeForce, RTX/Quadro/NVS, or Tesla lines below the fixed versions listed in bulletin 5861, e.g., GeForce < 616.56, RTX/Quadro/NVS < 616.92, Tesla < 616.92, plus multiple LTS branches). The attacker must ALREADY hold HIGH privileges (PR:H) to reach and exploit the kernel-mode driver interface; there is no network vector (AV:L) and no user interaction (UI:N). The chief limiting factors are the high pre-existing privilege requirement and local-only access, which together make remote or unprivileged exploitation impossible under the provided CVSS data. On GeForce, note one version branch (< 582.78) is scoped only to Maxwell, Volta, and Pascal architecture GPUs.

The CVSS 3.1 vector (AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H, base 6.7) describes a purely local vulnerability that additionally requires HIGH privileges (PR:H) to exploit - the attacker must already hold elevated/administrative access on the Linux host before triggering the kernel-mode out-of-bounds write. This significantly constrains real-world risk: an actor with PR:H already controls much of the system, so the primary incremental gain is kernel-level code execution from an already-privileged context (e.g., defeating kernel integrity/lockdown boundaries) rather than a bottom-up privilege escalation from an unprivileged user. No EPSS score, CISA KEV entry, or public POC is present in the provided data, so there is no evidence of active exploitation or automatable mass attack - this is no public exploit identified at time of analysis. The 'RCE' tag overstates reality: the vector is AV:L (local), not network, so remote code execution is not supported by the CVSS data. Net assessment: a legitimate kernel-mode memory-corruption bug worth patching on a normal cycle, but NOT an urgent/high-real-risk issue - the PR:H requirement and absence of exploitation signals keep priority low-to-medium despite the multi-impact CIA profile.

An attacker who has already obtained high-privilege (e.g., root or a privileged service) access on a Linux workstation or GPU compute node crafts malicious input to the NVIDIA kernel-mode driver interface, triggering an out-of-bounds write in kernel memory. By controlling the overwrite, they corrupt kernel structures to execute code at kernel privilege, potentially disabling kernel-integrity protections or reading/tampering with kernel memory. No public exploit code is identified, and exploitation requires pre-existing elevated local access.