Critical CVSS 9.8 RCE Flaw Discovered In ServiceNow AI Platform

CVE-2026-0542 Exposes Sandbox Component to Unauthenticated Remote Code Execution Risk

Published on Mar 2, 2026
Critical CVSS 9.8 RCE Flaw Discovered In ServiceNow AI Platform

A critical Remote Code Execution (RCE) flaw has been identified in ServiceNow’s AI Platform, exposing companies to potential unauthenticated compromise. Tracked as CVE-2026-0542, the flaw carries a CVSS severity score of 9.8.

According to ServiceNow’s security advisory, the vulnerability affects the platform’s sandbox component, which is intended to isolate untrusted code execution. Under certain conditions, an attacker could bypass these isolation controls and remotely execute arbitrary code over HTTPS without valid credentials.

RCE vulnerabilities are considered among the most severe classes of software flaws. Because they empower attackers to run commands within a target system, they can enable data theft, manipulation of automated workflows, and full system compromise.

Components affected by the vulnerability include web interfaces, Application Programming Interfaces (APIs), as well as automation modules in the AI software. Companies relying on ServiceNow to manage IT service management, security operations, and enterprise workflows, should be aware that such access could create significant operational and compliance risks.

ServiceNow stated it is not aware of active exploitation targeting customer instances at the time of disclosure. However, the company acknowledged the seriousness of an unauthenticated RCE in a core enterprise platform.

“This represents a huge risk, because AI agents inside ServiceNow can be very powerful, and perform actions such as create users and share data,” Melissa Ruzzi, director of AI at SaaS and AI security company AppOmni told Expert Insights.

“This is another example demonstrating how important it is to monitor AI and fully understand the controls in place for authentication and data flow within your AI platform,” she added.

On Jan. 6, 2026, the company deployed a security update to affected hosted customer instances. Fixes have also been made available to self-hosted customers and partners.

Patches span several releases. Zurich Patch 5 became available on Jan. 12, 2026, followed by Zurich Patch 4 Hotfix 3b on Feb. 23, 2026. Yokohama Patch 12 was issued on Feb. 6, 2026, with Patch 10 Hotfix 1b released on Feb. 18, 2026. Xanadu Patch 11 Hotfix 1a was published on Feb. 2, 2026. The Australia release is expected to receive a fix in Q2 2026.