
A critical remote code execution flaw in the ServiceNow AI platform tracked as CVE-2026-6875 is being exploited in the wild allowing unauthenticated attackers to run arbitrary code. ServiceNow has issued patches for its hosted instances while self hosted customers must apply the updates manually. ServiceNow's advisory details the issue.
The vulnerability scores 9.5 on the CVSS scale and stems from an oversight in the GlideRecord API where insufficient sandboxing permits user supplied input to be manipulated into a command execution vector. Affected releases include the Australia Patch 2 and Zurich Patch 7b builds among others. SecurityWeek reporting notes the flaw enables a sandbox escape without authentication.
Searchlight Cyber disclosed the flaw on July 14 2026 and published exploit details that were subsequently observed being used by the threat group Defused. Although ServiceNow initially reported no signs of active infection telemetry collected after disclosure shows the vulnerability is being leveraged in the wild to execute commands and access credentials. The Hacker News coverage highlights the pre authentication nature of the bug.
The emergence of this exploit adds to a series of unauthenticated remote code execution flaws affecting enterprise platforms this year highlighting the need for rapid patch management. No specific threat actor has been attributed to the observed attacks but the activity demonstrates that adversaries are quickly weaponizing newly disclosed vulnerabilities. SecurityAffairs article provides further analysis of the exploit chain.
Administrators should prioritize applying the latest ServiceNow patches to all hosted instances and ensure self hosted deployments receive the Australia Patch 2 or Zurich Patch 7b updates as appropriate. Reviewing system logs for unexpected GlideRecord queries and restricting direct internet exposure to the platform can help reduce risk. Implementing network segmentation and monitoring for atypical process creation are also recommended. SecurityOnline advisory offers a concise mitigation checklist.
Where possible disable unnecessary API endpoints that leverage GlideRecord enforce least privilege on service accounts and consider deploying a web application firewall rule set to block suspicious input patterns. Keeping abreast of future advisories from ServiceNow will ensure any further developments are addressed promptly. SecurityWeek roundup notes related patches from other vendors released the same day.