NVIDIA Triton Inference Server Stack Overflow Vulnerability Leading to Denial-of-Service

A stack-based buffer overflow vulnerability has been identified in NVIDIA Triton Inference Server for both Linux and Windows. This vulnerability allows an attacker to cause a stack overflow by sending excessively large payloads. Exploiting this vulnerability successfully could result in a denial-of-service condition.

Impact

Exploitation of this vulnerability can cause a denial-of-service condition, leading to a crash or unresponsiveness of the Triton Inference Server.

Remediation

Users are advised to update to version 25.09 or later. The latest release can be downloaded from the Triton Inference Server Releases page on GitHub. Additionally, consult the Secure Deployment Considerations Guide available on the same GitHub repository.