Threats Tagged 'cve-2026-44223'

Red Hat AI Inference

Red Hat AI Inference

Red Hat AI Inference

Red Hat AI Inference

CVE-2026-34753 is a server-side request forgery (SSRF) vulnerability found in the vLLM component of Red Hat AI Inference Server 3.4.4. It allows an attacker who can control batch input JSON to make arbitrary HTTP/HTTPS requests from the server, potentially accessing internal services such as cloud metadata endpoints or internal HTTP APIs. No official fix or patch is currently available from Red Hat for this vulnerability. The vulnerability has a Red Hat CVSS score of 5.4 (medium severity) but is assessed here as high severity due to the potential for information disclosure and further system compromise.

Red Hat AI Inference

The Requests library's utility function extract_zipped_paths() uses predictable filenames when extracting zip archive contents to the system temporary directory. If a file with the same name already exists, it is reused without validation, allowing a local attacker with write access to the temp directory to substitute a malicious file. Standard usage of the Requests library is not affected; only direct calls to extract_zipped_paths() are vulnerable. The issue is fixed in Requests version 2.33.0 by extracting files to a non-deterministic location. Alternatively, setting TMPDIR to a directory with restricted write access can mitigate the risk.

vLLM is an inference and serving engine for large language models (LLMs). From 0.18.0 to before 0.20.0, the extract_hidden_states speculative decoding proposer in vLLM returns a tensor with an incorrect shape after the first decode step, causing a RuntimeError that crashes the EngineCore process. The crash is triggered when any request in the batch uses sampling penalty parameters (repetition_penalty, frequency_penalty, or presence_penalty). A single request with a penalty parameter (e.g., "repetition_penalty": 1.1) is sufficient to crash the server. This vulnerability is fixed in 0.20.0.