CVE-2026-110220

Awaiting Analysis Awaiting Analysis - Queue

Server-Side Request Forgery in Kubeflow Model Registry

Publication date: 2026-10-13

Last updated on: 2026-10-13

Assigner: VulnCheck

Description

Kubeflow Model Registry through 0.3.17 contains a server-side request forgery vulnerability that allows namespace users to steal the controller ServiceAccount token via the modelregistry.kubeflow.org/url annotation. Attackers can create an InferenceService pointing the annotation at their server, capture the bearer token, and modify other tenants' InferenceServices cluster-wide.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published

2026-10-13

Last Modified

2026-10-13

Generated

2026-10-19

AI Q&A

2026-10-14

EPSS Evaluated

2026-10-18

NVD

CVE-2026-110220

EUVD

EUVD-2026-99261

Affected Vendors & Products

Vendor	Product	Version / Range
Kubeflow	Model	Registry 0

Helpful Resources

Exploitability

CWE

KEV

CWE ID	Description
CWE-918	The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination.

Attack-Flow Graph

Executive Summary

This is a Server-Side Request Forgery (SSRF) vulnerability in Kubeflow Model Registry through version 0.3.17. It allows namespace users to steal the controller's Kubernetes ServiceAccount token by exploiting the modelregistry.kubeflow.org/url annotation in an InferenceService. Attackers can create an InferenceService with a malicious URL pointing to their server, causing the controller to send its bearer token to the attacker's host. This token can then be used to modify other tenants' InferenceServices cluster-wide.

Detection Guidance

To detect this vulnerability, inspect InferenceService resources for the modelregistry.kubeflow.org/url annotation pointing to external or suspicious hosts. Use kubectl get inferenceservices --all-namespaces -o yaml to list all InferenceServices and check for unauthorized URL annotations. Monitor controller logs for unexpected outbound connections to untrusted destinations.

Impact Analysis

If exploited, this vulnerability allows attackers to capture the controller's ServiceAccount token, which typically has cluster-wide privileges. This can lead to unauthorized access to other tenants' InferenceServices, potential data breaches, or further compromise of the Kubernetes cluster. The impact includes credential theft, unauthorized modifications, and potential denial-of-service conditions.

Compliance Impact

This vulnerability could lead to unauthorized access to sensitive data, violating GDPR's data protection requirements and HIPAA's security rules. It may result in data breaches, unauthorized data exposure, or loss of control over personal or health information, potentially leading to regulatory penalties and compliance violations.

Mitigation Strategies

Immediately upgrade to a patched version of Kubeflow Model Registry if available. If not, restrict permissions to prevent namespace users from creating or modifying InferenceService resources. Remove or validate the modelregistry.kubeflow.org/url annotation in all InferenceService objects. Audit and revoke any exposed ServiceAccount tokens.

Hi! I’m here to help you understand CVE-2026-110220. Ask me anything about the vulnerability, its impact, or mitigation strategies.

0/70