GitLab warns of critical RCE vulnerability in AI Gateway service

GitLab warns of critical RCE vulnerability in AI Gateway service
GitLab has warned customers to urgently patch CVE-2026-90970, a critical AI Gateway vulnerability that could let authenticated attackers with basic privileges escape the sandbox and execute arbitrary commands on vulnerable self-hosted instances. The company released fixed versions for GitLab Self-Hosted AI Gateway users and said hosted GitLab AI Gateway customers are already protected, following its recent patching of CVE-2026-85706. #GitLab #CVE-2026-90970 #CVE-2026-85706 #CISA

Keypoints

  • GitLab says CVE-2026-90970 can lead to arbitrary command execution on AI Gateway.
  • The flaw affects GitLab Self-Hosted AI Gateway deployments.
  • Attackers need basic privileges and Duo Agent Platform access to abuse it.
  • GitLab released versions 19.2.4, 19.3.2, and 19.4.1 to fix the issue.
  • GitLab also recently patched CVE-2026-85706, a path traversal flaw added to CISA’s actively exploited list.

Read More: https://www.bleepingcomputer.com/news/security/gitlab-warns-of-critical-rce-vulnerability-in-ai-gateway-service/