GitLab Warns of Critical RCE Vulnerability in AI Gateway Service

What Happened

GitLab has issued an advisory urging customers to immediately patch a critical remote code execution (RCE) vulnerability affecting its AI Gateway service, according to BleepingComputer. The flaw could allow attackers to run arbitrary commands on vulnerable instances. At the time of this writing, GitLab has not published a CVE identifier, patch version numbers, or technical exploitation details in the reporting available to us, so specifics of the affected versions and the exact attack vector remain unconfirmed.

Why It Matters

AI Gateway is the component that brokers requests between GitLab and AI-assisted features (such as Duo). A critical RCE in this service is significant because it sits in a privileged position within the GitLab architecture — self-managed GitLab instances that have the AI Gateway service deployed are the most likely to be exposed, and successful exploitation could give an attacker a foothold inside source code management infrastructure, which is a high-value target for supply chain and credential-theft attacks.

What Defenders Should Do Now

• Identify whether your organization runs a self-managed GitLab AI Gateway deployment and check GitLab's official security advisories for the specific version and patch guidance.
• Prioritize patching AI Gateway instances as GitLab releases fix details — treat this as an urgent, out-of-cycle update given the "critical" and RCE classification.
• In the interim, review network exposure of the AI Gateway service (restrict access to trusted networks/VPN where possible) and audit logs for unexpected process execution or outbound connections from hosts running the service.
• Once GitLab publishes IOCs or technical details, review GitLab and system logs around the AI Gateway for anomalous command execution, unexpected child processes, or unusual outbound traffic consistent with post-exploitation activity.

Developing Story

This is based on an initial vendor advisory report and details are still emerging — no CVE ID, affected version range, or proof-of-concept has been confirmed in the source material reviewed. We will track this story and update our detection content as GitLab releases further technical guidance. Read the original report at BleepingComputer.