A critical vulnerability in the ServiceNow AI Platform, identified as CVE-2026-6875, is currently being actively exploited by attackers. This flaw allows unauthenticated individuals to escape the platform’s script sandbox and execute arbitrary code on affected systems.
ServiceNow’s AI Platform is a widely used Platform-as-a-Service (PaaS) that enables organizations to develop and automate digital workflows. The vulnerability, a code injection issue, permits attackers without authentication to bypass the script sandbox and execute code remotely on targeted instances. This flaw was discovered by researchers at Searchlight Cyber and reported to ServiceNow on April 1, 2026. ServiceNow promptly addressed the issue by deploying security updates to hosted instances and providing patches for self-hosted customers and partners throughout June. The existence of CVE-2026-6875 was publicly disclosed on July 13, 2026.
Despite these proactive measures, threat intelligence firm Defused has confirmed that attackers began exploiting this vulnerability in the wild starting July 18, 2026. Notably, these malicious actors have employed multiple sandbox-escape techniques, some differing from the initial proof-of-concept documented by Searchlight Cyber. This indicates that organizations relying solely on known mitigation strategies may remain vulnerable.
ServiceNow has issued security updates to address this critical flaw. Hosted customers have received automatic updates, while patches have been made available for self-hosted deployments and partners. Organizations utilizing self-managed ServiceNow instances are strongly advised to apply the latest security updates or upgrade to a patched release immediately. Additionally, enabling Guarded Script is recommended to mitigate sandbox escape attacks by restricting certain JavaScript expressions in sandboxed contexts.
Security teams should take the following actions to safeguard their systems:
- Review the status of ServiceNow updates to ensure all patches have been applied.
- Inspect logs for suspicious requests targeting the `/assessment_thanks.do` endpoint.
- Investigate unusual parameters associated with `sysparm_assessable_type`.
- Monitor for unexpected administrative account creations and abnormal script activity.
- Scrutinize actions involving MID Servers for any irregularities.
The active exploitation of CVE-2026-6875 underscores the critical importance of timely patch management and vigilant monitoring. Organizations must prioritize the application of security updates and remain alert to emerging threats to protect their systems and sensitive data from unauthorized access and potential breaches.
