---
id: "CVE-2026-17557"
kind: "vulnerability"
aliases: ["CVE-2026-17557"]
title: "Red Hat OpenShift AI: evaluation job leaks model secret and enables SSRF"
severity: "medium"
cvss: 6.5
epss: 0.00258
exploited: false
status: "open"
category: "ai-tooling"
products: []
ecosystems: []
tags: ["ai-tooling", "kubernetes", "lm-evaluation-harness", "multi-tenant", "openshift-ai"]
not_affected_if: ["no-evaluation-jobs"]
published_at: "2026-10-21T08:16:43+00:00"
first_seen_at: "2026-10-23T21:57:28+00:00"
updated_at: "2026-10-25T03:49:05+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-17557", "first_seen_at": "2026-10-23T21:57:28+00:00"}, {"source": "epss", "url": "https://api.first.org/data/v1/epss?cve=CVE-2026-17557", "first_seen_at": "2026-10-24T13:15:54+00:00"}]
---

# Red Hat OpenShift AI: evaluation job leaks model secret and enables SSRF

An authenticated user who submits an evaluation job can point its model URL at an external listener, causing the tenant's model authentication secret to be sent there. The same job can also be used to make server-side requests to internal network endpoints reachable from the evaluation pod.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- unknown

## Fix

- unknown

## Timeline

- 2026-10-23T21:57:28+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-17557
- 2026-10-24T13:15:54+00:00 epss https://api.first.org/data/v1/epss?cve=CVE-2026-17557

## From the sources

A flaw was found in the `lm-evaluation-harness` component of Red Hat OpenShift AI. An authenticated user, able to submit an evaluation job, could exploit this by providing a malicious model URL in the `JobSpec`. This vulnerability allows for the disclosure of the tenant's model authentication secret to an attacker-controlled external listener. Additionally, it enables Server-Side Request Forgery (SSRF) attacks against internal network endpoints reachable from the evaluation job pod.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_