ServiceNow patches AI Platform flaw allowing host code execution

ServiceNow issued security updates on 29 October 2026 for three flaws in its cloud ecosystem. The most severe, CVE-2026-103093, is rated 9.4 on CVSSv4 and affects the ServiceNow AI Platform. ServiceNow says an authenticated user with the admin role could bypass Rhino sandbox restrictions and run arbitrary code on the underlying host operating system, potentially gaining further unintended access. The flaw was identified through internal testing and bug bounty reports.

CVE-2026-103094, rated 7.1, could cause a user’s plaintext password to be stored in a system table, where someone not entitled to see it might gain access. CVE-2026-6877, rated 5.3, affects the SR – CSM – Task Plan Templates application: authenticated users with specific roles could create records across domain boundaries. The article says the first two issues affect AI Platform releases in the Yokohama, Zurich and Australia families; CVE-2026-6877 affects Task Plan Templates versions 2.0.0 and earlier.

ServiceNow said it had found no evidence of malicious exploitation, and the article reported no public proof-of-concept code. For the AI Platform flaws, fixes include Yokohama Patch 13 Hotfix 4a/5a, Zurich Patch 10 Hotfix 3b/4b and Australia Patch 5. Task Plan Templates customers should upgrade to version 3.0.0 or later, or install version 2.0.1 through recent Australia patches.