AI Identity Theft: Critical ServiceNow Flaw (CVE-2025-12420) Allows Unauthenticated Impersonation

A critical vulnerability in the ServiceNow AI Platform, tracked as CVE-2025-12420, could allow unauthenticated attackers to masquerade as legitimate users, potentially enabling privilege escalation. With a severity of 9.3 out of 10, the flaw centres on a failure in authentication checks that could let an unauthenticated user impersonate another user and perform that user’s authorised operations.

The vulnerability was disclosed in October 2025 by AppOmni, with credit also going to researcher Aaron Costello for their collaboration. ServiceNow released security updates for most hosted instances on 30 October 2025, though self-hosted customers and partners are urged to take immediate action to secure their networks.

Updates have been released for specific Store Applications: Now Assist AI Agents (sn_aia) to version 5.1.18 or later (or 5.2.19 or later), and Virtual Agent API (sn_va_as_service) to version 3.15.2 or later (or 4.0.4 or later). The article, dated 13 January 2026, notes that ServiceNow is unaware of active exploitation in the wild but emphasises vigilance given the potential risk when vulnerabilities are disclosed.