CVE-2026-12491

MEDIUMCVSS 4.8/10EPSS 0.24%

Last modified

CVE-2026-12491 is a medium-severity vulnerability rated 4.8/10 on the CVSS scale. 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. EPSS estimates a 0.24% chance of exploitation in the next 30 days.

Description

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.

Metrics

CVSS 3.1
4.8/10

CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:L

EPSS Probability
0.24%

14.9th percentile

Probability of exploitation in the next 30 days. Learn more

Weakness Enumeration

Affected Software

Source: CNA advisory (CVE.org). NVD analysis pending.

VendorProductVersions
vllm-projectvLLM>= 0.11.0, < 0.24.0
Red HatRed Hat AI Inference ServerAll versions
Red HatRed Hat Enterprise Linux AI (RHEL AI) 3All versions
Red HatRed Hat OpenShift AI (RHOAI)All versions

References

Timeline

Published
Last Modified
Status
Awaiting Analysis

Frequently Asked Questions

What is CVE-2026-12491?
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.
How severe is CVE-2026-12491?
CVE-2026-12491 has a CVSS score of 4.8/10 (MEDIUM severity). The EPSS model estimates a 0.24% probability of exploitation in the next 30 days.
How do I fix CVE-2026-12491?
Check the vendor references and advisories linked above for patched versions and mitigation guidance. You can also run a Strix scan to test if your systems are affected.

How Strix Helps

Related CVEs from 2026

Are you affected by CVE-2026-12491?

Run a free Strix scan to check your systems for this vulnerability.

Scan your code now

Source: NVD / NIST