ServiceNow AI Platform CVE-2026-103094

Exploitation requires (1) a valid, authenticated low-privileged ServiceNow AI Platform account (PR:L), and (2) that a user's plaintext password was actually unintentionally persisted into the affected system table - the AT:P metric reflects that this conditional persistence must have occurred, so it is not guaranteed in every instance. The attacker must be able to view/query that system table. What LIMITS exploitation: authentication is mandatory (no anonymous access), impact is confined to instances where the plaintext persistence actually happened, and it is resolved by upgrading to a fixed release (e.g., Australia Patch 4 Hot Fix 1 / Patch 4m / Patch 5, Zurich Patch 11 Hot Fix 1 / Patch 12 / Patch 13, Yokohama Patch 13 Hot Fix 4a, Xanadu Patch 11 Hot Fix 7a, per KB3228994). No user interaction by the victim is required (UI:N).

This is a credible but not emergency-grade priority, consistent with the vuln.today 'this_week' decision label. The CVSS 4.0 vector (AV:N/AC:L/AT:P/PR:L/UI:N) shows a network-reachable flaw of low complexity, but it is NOT unauthenticated: PR:L means an attacker must already hold a valid low-privileged ServiceNow account, and AT:P (Attack Requirements: Present) indicates exploitation depends on a precondition outside the attacker's direct control - namely that a user's plaintext password actually got persisted into the system table. The impact side is what drives the 7.1 score: VC:H (the disclosed plaintext credential is high-value confidentiality loss) plus subsequent-system impact (SC:H/SI:H), reflecting that a stolen plaintext password can be reused to access or alter resources beyond the originally authorized scope. No EPSS score, KEV listing, or PoC/exploit reference was provided in the input; ServiceNow explicitly states it is not currently aware of malicious exploitation. There is therefore no evidence of active exploitation or weaponization - the urgency here comes from the vendor advisory and decision label, not from exploitation telemetry. Net: a real, patch-this-week issue because it leaks credentials and enables lateral/privilege reuse, but gated behind authentication and a conditional trigger, so it is not a drop-everything unauthenticated RCE.

An authenticated low-privileged ServiceNow user browses or queries a system table that, due to the flaw, retained another user's password in plaintext, and reads a credential they were never entitled to see. They then reuse that plaintext password to log in as the victim and access data or functions beyond their own authorization. No public exploit identified at time of analysis, and ServiceNow reports no known malicious exploitation.