ServiceNow — Critical Maximum-Severity Flaws Allow Unauthenticated Remote Code Execution and SQLi (CVE-2026-18885, CVE-2026-18886, CVE-2026-74820)

Publication date: August 28, 2026
Category: Vulnerability / Cloud Security

Introduction

ServiceNow has released urgent security updates to address four critical vulnerabilities impacting the ServiceNow AI Platform. Among the resolved issues, three have been rated at the maximum severity score of 10.0 on the CVSS v4.0 scale. These flaws can be exploited, under certain conditions, by unauthenticated remote attackers to compromise the integrity and confidentiality of enterprise environments. The vulnerabilities involve code injection within GraphQL APIs, access control flaws in configuration image upload processors, and SQL injection vulnerabilities via dynamic schema ORDER BY clauses.

What is the ServiceNow Platform and Its Security Flaws? (General Analysis)

ServiceNow is a cloud-based enterprise operating system that automates workflows, IT service management (ITSM), and artificial intelligence operations for global organizations. Its core position in corporate IT infrastructure makes it a high-value target for threat actors.

This patch cycle addresses the following official CVEs verified by NVD:

  • CVE-2026-18885: A code injection vulnerability (CWE-94) in the GraphQL Composite Data API, rated with a CVSS v4.0 score of 10.0 (Critical).
  • CVE-2026-18886: An improper access control vulnerability (CWE-284) in the system configuration image upload processor, rated with a CVSS v4.0 score of 10.0 (Critical).
  • CVE-2026-74820: A SQL injection vulnerability (CWE-89) reached through a dynamic schema ORDER BY clause, rated with a CVSS v4.0 score of 10.0 (Critical).
  • CVE-2026-6876: A sandbox escape (CWE-94, CWE-693, CWE-1284) in the Now Platform, rated with a CVSS v4.0 score of 8.7 (High).
  • CVE-2026-6875: A remote code execution vulnerability in the AI platform, rated with a CVSS v4.0 score of 9.5 (Critical).

How Does It Work? (Technical Analysis)

Analysis of the attack vectors highlights a combination of architectural weaknesses in user input processing and privilege segregation within the platform engine:

  • GraphQL Input and Processing Flow (CVE-2026-18885): Requests directed to the GraphQL Composite Data API fail to properly validate or sanitize complex data structures. An external attacker can craft malicious queries to inject and execute arbitrary code within the application server context.
  • Access Control Manipulation via Image Processor (CVE-2026-18886): The system configuration image upload processor lacks robust pre-authentication restrictions. This allows unauthorized users to create or modify instance records, effectively resulting in privilege escalation.
  • SQL Injection via Dynamic Clauses (CVE-2026-74820): When processing sorting operations through the dynamic ORDER BY schema parameter, the platform concatenates untrusted user input directly into SQL statements executed against the underlying database.
  • Sandbox Escapes and Restricted Contexts: Prior research into AI components and the platform (CVE-2026-6875 and CVE-2026-6876) has shown that logical sandbox boundaries can be bypassed if foundational logic flaws exist, enabling system- or platform-level command execution.

Affected Systems / Environments

The vulnerabilities impact multiple ServiceNow product families and versions, requiring specific updates depending on the deployment model:

CVECategory (CWE)ImpactCVSSVector (summary)
CVE-2026-18885CWE-94 (Code Injection)Critical10.0AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H
CVE-2026-18886CWE-284 (Improper Access Control)Critical10.0AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H
CVE-2026-74820CWE-89 (SQL Injection)Critical10.0AV:N/AC:L/AT:N/PR:N/VC:H/VI:H/VA:H
CVE-2026-6876CWE-94, CWE-693 (Sandbox Escape)High8.7AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H

Broadly affected family versions include Xanadu (prior to Patch 11 Hot Fix 7a), Yokohama (prior to Patch 12 Hot Fix 3b and Patch 13 Hot Fix 4), Zurich, and Australia, as specified in the manufacturer’s advisory.

Mitigation and Detection

Remediation

  • Immediate Patching: ServiceNow has deployed automated security updates to hosted cloud instances; however, self-hosted (on-premise or private cloud) customers and partners must urgently apply the hotfixes and patches listed in the August advisory.
  • Version Verification: Ensure that all deployments across Xanadu, Yokohama, Zurich, and Australia meet or exceed the minimum patched release levels specified by the vendor.

Detection

  • Monitor platform access and error logs for anomalous requests targeting GraphQL endpoints and image upload utilities.
  • Deploy Web Application Firewall (WAF) rules to inspect suspicious query strings attempting to manipulate ORDER BY clauses and code injection patterns.

Defensive Intelligence Note: The pre-authenticated nature of the 10.0-rated flaws means that any public exposure of unpatched instances can be weaponized remotely without requiring prior valid credentials.

Wrapping Up

The disclosure of three CVSS 10.0 vulnerabilities in ServiceNow highlights the critical attack surface represented by modern enterprise platforms driven by AI and complex APIs. While ServiceNow has automated mitigations for hosted instances, self-hosted administrators bear the immediate responsibility of applying patches to secure their organizations’ operational integrity.

References