External risk intelligence

NVIDIA TensorRT-LLM allows attackers to take control, change data, or steal information.

CVE advisorySeverity: CRITICAL (CVSS 9.8)

CVE-2026-24142

NVIDIA TRT-LLM is an AI inference library integrated into backend services. While these services may be internet-facing, the library acts as a component rather than a standalone edge or gateway service, making internet exposure dependent on the specific application implementation.

Deserialization

Nvidia Tensorrt Llm

before 1.2

Halo Surface Signal: 3 out of 5 — possibly public-facing.

External exposure likelihood

Horizon Alert

Summary of the vulnerability and why it matters

NVIDIA TRT-LLM has a deserialization vulnerability that could allow attackers to execute code, alter data, or access sensitive information. This issue is critical because it can be exploited remotely without any required privileges.

Remote code execution is possible.
Data integrity and confidentiality are at risk.
Unauthenticated access can lead to compromise.

Attack Path

How an attacker could exploit the issue

An unauthenticated attacker could exploit this vulnerability by sending specially crafted serialized data to a vulnerable instance of NVIDIA TRT-LLM. This could allow them to execute arbitrary code on the system, leading to unauthorized data modification or the exposure of sensitive information.

Network accessible
Unsafe deserialization
No user interaction needed

Live Threat

Current exploitation, exposure, and threat context

This vulnerability in NVIDIA TRT-LLM, allowing code execution and data tampering, presents a significant potential threat. Because it affects a component used in backend AI services, exploitation would likely require specific knowledge of the target's infrastructure. However, the critical severity and broad impact make it a tempting target for sophisticated attackers.

Likely targeted by advanced adversaries.
Exploitability depends on service implementation.
No public exploit code currently observed.

Operational Fix

Recommended remediation, mitigation, and detection steps

Prioritize isolating or taking offline any services using NVIDIA TensorRT-LLM versions prior to 1.2 due to a critical deserialization vulnerability that can lead to code execution, data tampering, and information disclosure. Given the CVSS score of 9.8 and potential for full system compromise, immediate action is required if affected services are exposed externally or if there's evidence of active exploitation.

Patch NVIDIA TensorRT-LLM to version 1.2 or later.
Isolate affected systems from untrusted networks.
Monitor network traffic for exploitation indicators.

Supplementary metadata

CVSS vector

Validate whether this threat affects your internet-facing exposure.

Halo Threat Intelligence helps prioritize remediation with Halo Surface Signal and H/A/L/O context. Start exposure validation with a free external attack surface trial.

Free EASM Trial See How Signal Works

Frequently asked questions

What is the software context for CVE-2026-24142, affecting NVIDIA TensorRT-LLM?

CVE-2026-24142 impacts NVIDIA TensorRT-LLM, an AI inference library. This vulnerability exists in versions prior to 1.2 and could allow for code execution, data tampering, and information disclosure due to a deserialization flaw.

How is the deserialization vulnerability in NVIDIA TensorRT-LLM decoded, and what is its weakness class?

The vulnerability in NVIDIA TensorRT-LLM is a deserialization flaw (CWE-502). This weakness allows a successful exploit to lead to code execution, data tampering, and information disclosure when processing unsafe serialized data.

What is the trigger path and scope for the NVIDIA TensorRT-LLM vulnerability?

An unauthenticated attacker can trigger this vulnerability by sending specially crafted serialized data to a vulnerable instance of NVIDIA TensorRT-LLM. The scope is not modified (S:U), meaning the vulnerability impacts the same security authority.

What is the relevance of CVE-2026-24142 concerning the Halo Surface Signal?

The Halo Surface Signal indicates a 'Possible' threat for CVE-2026-24142. While NVIDIA TRT-LLM is an AI inference library used in backend services, its internet exposure is dependent on the specific application implementation, making it a component rather than a standalone edge service.

What practical steps should be taken to address the NVIDIA TensorRT-LLM vulnerability?

To address CVE-2026-24142, users should patch NVIDIA TensorRT-LLM to version 1.2 or later. It is also recommended to isolate affected systems from untrusted networks and monitor network traffic for any signs of exploitation.

References

Written by Nicholas Merritt

Nicholas Merritt is VP of Security Solutions at Halo Security. With more than 25 years across application security, vulnerability prioritization, and offensive security, he helps teams translate threat data into practical exposure validation and remediation focus. He authors Halo Threat Intelligence CVE advisories using Halo Surface Signal and H/A/L/O context to prioritize internet-facing risk.