GitLab AI Gateway

GitLab has fixed a critical security flaw tracked as CVE-2026-90970 affecting its AI Gateway component. Security researchers urge organizations running the platform to apply the patch promptly, as critical flaws in AI infrastructure can expose sensitive data and models to attack. No details on active exploitation were provided, but administrators are advised to update their installations as soon as possible.

GitLab has released emergency security patches addressing two critical vulnerabilities: a remote code execution flaw in its AI Gateway, tracked as CVE-2026-90970, and a maximum-severity path traversal issue. Both flaws are reportedly being actively exploited, prompting the unusually urgent rollout. Security teams are being urged to apply the updates immediately, with the disclosures fueling renewed discussion about securing AI infrastructure.

GitLab has patched a critical vulnerability, CVE-2026-90970, rated CVSS 9.9, in its AI Gateway. The flaw allows an authenticated attacker to escape the prompt-template sandbox and execute arbitrary commands on self-hosted deployments. Security professionals are urging administrators to apply the update quickly and check whether their installations are affected.

GitLab has released fixes for CVE-2026-90970, a critical vulnerability (CVSS 9.9) in its AI Gateway. An authenticated user with access to the Duo Agent Platform can run commands on the gateway. Only self-hosted gateway deployments are affected. Patches are available in versions 19.2.4, 19.3.2 and 19.4.1, and administrators are urged to update immediately.

A security vulnerability tracked as CVE-2026-90970 has been disclosed in GitLab's AI Gateway, reportedly allowing remote code execution on affected systems. GitLab's AI Gateway sits in front of the company's AI-powered features, so a flaw there could expose organizations using the platform's AI tooling. Administrators are being urged to check their deployments and apply patches. Details on affected versions and exploitation are still limited.