A critical authorization vulnerability in GitLab EE allows authenticated users to access restricted AI model configurations, creating a significant risk of unauthorized data manipulation.

The software suffers from a missing authorization check (CWE-862) within its API, which permits an authenticated user to perform unauthorized actions on AI model settings by manipulating namespace identifiers.

This vulnerability poses a substantial risk to organizations leveraging AI features within GitLab, as it allows internal users to bypass access controls and interact with configurations they are not authorized to manage. Given the CVSS score of 7.1, this is a high-severity issue that could lead to the exposure or corruption of proprietary AI models and sensitive configuration data. Unauthorized access to these settings can disrupt development workflows and potentially compromise the integrity of AI-assisted features.

Immediate Action: Upgrade all GitLab EE instances to version 18.5.5, 18.6.3, 18.7.1, or later to incorporate the necessary authorization checks.

Proactive Monitoring: Review API access logs for anomalous requests targeting namespace identifiers or AI model endpoints, particularly those originating from accounts with restricted permissions.

Compensating Controls: Implement strict API rate limiting and ensure that Web Application Firewalls (WAF) are configured to monitor for suspicious patterns in API calls that involve namespace manipulation.

Public Exploit Available: No

The risk posed by this authorization bypass warrants immediate remediation. Organizations should prioritize patching their GitLab environments to the recommended versions to prevent unauthorized users from tampering with AI model settings. Regular audits of API usage and adherence to the principle of least privilege for all authenticated users remain essential to defending against similar authorization flaws.