Privilege Escalation Vulnerability in ServiceNow AI Platform

Du'aine Davis

For awareness

• October 2025, AppOmni disclosed a critical privilege‑escalation flaw (CVE‑2025‑12420) affecting the ServiceNow AI Platform, allowing unauthenticated attackers to impersonate any user. ServiceNow issued patches on 30 October for most hosted instances and provided updates for partners and self‑hosted deployments. As of January 2026, there is no evidence of exploitation in the wild.
• The vulnerability enabled an unauthenticated actor to exploit a flaw in the AI Platform’s impersonation logic, granting them the ability to perform actions permitted to the impersonated user. AppOmni responsibly disclosed the issue, and the advisory does not attribute exploitation to any threat group or malicious actor. The risk increased after public disclosure, prompting the advisory.
• Organisations must apply the fixed ServiceNow Store App versions (Now Assist AI Agents ≥5.1.18/5.2.19; Virtual Agent API ≥3.15.2/4.0.4). Defender security teams should monitor for unexpected ServiceNow API calls, anomalous impersonation events, and unauthenticated session attempts. Behaviour‑based detections should flag identity‑related anomalies such as privilege escalation attempts, unusual automation patterns, or API actions executed without user authentication.

[Security Advisory] CVE-2025-12420 - Privilege Escalation in ServiceNow AI Platform - Security
NVD - CVE-2025-12420