Technical impact
The vulnerability in NVIDIA TensorRT‑LLM is a server‑side request forgery flaw that allows an attacker to manipulate the application into making HTTP requests to arbitrary internal or external addresses. By submitting a crafted URL or request payload, the attacker can force the server to access internal services, retrieve sensitive data, or modify internal state, potentially leading to the disclosure of confidential information or tampering with data processed by the model. This weakness is identified as CWE‑918.
Affected Systems
The affected product is NVIDIA TensorRT‑LLM. No specific version range is provided in the available data, so any deployment of any release of TensorRT‑LLM should be considered potentially vulnerable until a vendor‑supplied fix is applied.
Risk and Exploitability
The CVSS score of 7.1 indicates a high severity. The EPSS score is not available and the vulnerability is not listed in the CISA KEV catalog. Because the flaw relies on the ability to submit an external URL or request, the likely attack vector is remote exploitation via the model’s input or API. Exploitation requires network connectivity to the vulnerable service and the ability to influence request parameters. If successful, the attacker can read internal data or alter data being processed by the TensorRT‑LLM instance.