NVIDIA TensorRT-LLM Deserialization Code Execution

Executive Summary

This vulnerability exists in NVIDIA TensorRT-LLM for Linux due to a flaw in the restricted unpickler used for model weight deserialization. It allows a local unauthenticated attacker to deserialize untrusted data, potentially leading to code execution, privilege escalation, data tampering, or information disclosure.

Useful 0 Not Useful 0

Detection Guidance

Detection of this vulnerability requires checking for the presence of NVIDIA TensorRT-LLM on your system and verifying if it is affected by the deserialization issue. Inspect installed packages using commands like 'dpkg -l | grep tensorrt-llm' on Debian-based systems or 'rpm -qa | grep tensorrt-llm' on RPM-based systems. Ensure the software is updated to the latest patched version.

Useful 0 Not Useful 0

Impact Analysis

If exploited, this vulnerability could allow attackers to execute arbitrary code on your system, gain higher privileges, modify or delete data, or steal sensitive information. It primarily affects systems running NVIDIA TensorRT-LLM for Linux.

Useful 0 Not Useful 0

Compliance Impact

The vulnerability could lead to data tampering and information disclosure, which may violate GDPR's data integrity and confidentiality requirements. For HIPAA, unauthorized code execution or privilege escalation could compromise protected health information.

Useful 0 Not Useful 0

Mitigation Strategies

Update NVIDIA TensorRT-LLM to the latest patched version immediately. Restrict access to systems running TensorRT-LLM to trusted users only. Monitor for unusual activity or unauthorized access attempts.

Useful 0 Not Useful 0