Critical ServiceNow AI Platform Sandbox Escape Enables Unauthenticated RCE
Emerging Technology Risk · 22 sources · Updated 4mo ago · First seen Jul 13, 2026

ServiceNow disclosed and remediated CVE-2026-6875, a critical sandbox escape in its AI Platform that can allow unauthenticated remote code execution in certain circumstances. The vulnerability affects both hosted and self-hosted ServiceNow environments, carries a CVSS 4.0 score of 9.5, and was published in advisory KB3137947. ServiceNow said it has already deployed security updates to hosted instances and issued patched family releases and updates for self-hosted customers and partners.
The company said it is not currently aware of active exploitation, but warned that successful compromise could let attackers execute code within the ServiceNow platform, disrupt workflows, access sensitive data, modify records, and enable follow-on malicious activity. To reduce near-term exploit risk, ServiceNow limited public technical details while directing customers to promptly apply the relevant updates or upgrade to patched releases, including fixes for the Brazil, Australia, Zurich, and Yokohama release families.


