CVE-2026-17557: Server-Side Request Forgery (SSRF) in Red Hat Red Hat OpenShift AI (RHOAI)

The vulnerability exists in the lm-evaluation-harness component of Red Hat OpenShift AI. It allows an authenticated user with permission to submit evaluation jobs to provide a crafted model URL in the JobSpec. Exploiting this flaw can lead to disclosure of the tenant's model authentication secret to an external attacker-controlled listener. Additionally, it enables SSRF attacks targeting internal network endpoints reachable from the evaluation job pod, potentially allowing attackers to interact with internal services indirectly.

Successful exploitation results in the disclosure of sensitive authentication secrets related to the tenant's model, which could be used by attackers to impersonate or misuse the model. Furthermore, SSRF attacks could allow attackers to probe or interact with internal network endpoints that are otherwise inaccessible, potentially leading to further internal reconnaissance or exploitation. There is no indication of direct integrity or availability impact from this vulnerability.

The vendor advisory from Red Hat should be consulted for the latest remediation guidance. No explicit patch or fix information is provided in the input data. Patch status is not yet confirmed — check the Red Hat advisory at https://access.redhat.com/security/cve/CVE-2026-17557 for current remediation details. Until a fix is applied, restrict authenticated user permissions to submit evaluation jobs only to trusted users and monitor usage closely.