PT-2026-50278 · Red Hat · Red Hat Ai Inference Server+2

Published 2026-06-17 · Updated 2026-06-17 · CVE-2026-12491

Mitre

CVSS v3.1

4.8

Medium

Base vector

AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:L

A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from improper handling of image metadata, specifically EXIF orientation and PNG transparency (tRNS) data, during image processing. When images are converted to RGB, transparency information may be implicitly discarded or remapped, leading to unexpected rendering of transparent pixels and distortion of input content. This can result in the model misinterpreting image content, potentially affecting the integrity of processed data.

Fix

Weakness Enumeration

CWE-115

Related Identifiers

CVE-2026-12491

Affected Products

Red Hat Ai Inference Server

Red Hat Enterprise Linux Ai (Rhel Ai) 3

Red Hat Openshift

References · 2

• https://bugzilla.redhat.com/show_bug.cgi?id=2489786 · Note
• https://access.redhat.com/security/cve/CVE-2026-12491 · Note