Attackers Exploit CVE-2026-6875 in ServiceNow AI Platforms
Threat actors are actively exploiting a critical CVSS 9.5 pre-authentication remote code execution vulnerability in self-hosted ServiceNow instances.
Threat actors are actively exploiting CVE-2026-6875, a critical code injection vulnerability in the ServiceNow AI Platform. According to recent tracking by cybersecurity firm Defused, active abuse began on July 17, 2026. The vulnerability carries a CVSS v4.0 score of 9.5 and grants unauthenticated attackers full remote code execution on targeted systems.
ServiceNow patched its hosted cloud infrastructure shortly after the initial vulnerability report on April 1, 2026. Self-hosted and on-premises customers did not receive security updates until July 13. This delay created a wide exposure window for enterprise instances operating outside the managed cloud boundary.
Sandbox Escape Mechanics
The vulnerability allows attackers to escape the ServiceNow script sandbox. By targeting the pre-authentication /assessment_thanks.do endpoint, malicious payloads bypass standard access controls to execute arbitrary code on the underlying instance and connected proxy servers.
While security researcher Adam Kues published a specific proof-of-concept demonstrating one path to execution, attackers have adapted quickly. Current in-the-wild exploitation relies on a different sandbox-escape gadget to achieve the same execution primitive. This mirrors complex attack patterns seen in other recent AI exploit chains, where threat actors modify execution paths immediately following public disclosure to bypass preliminary detection rules.
Incident and Remediation Timeline
| Event | Date | Status |
|---|---|---|
| Initial Discovery | April 1, 2026 | Reported by Searchlight Cyber |
| Cloud Remediation | April 2, 2026 | Hosted instances patched silently |
| Public Disclosure | July 13, 2026 | Patch released for on-premises deployments |
| Active Exploitation | July 17, 2026 | First observed by Defused in the wild |
Guarded Script and Enterprise Risk
The July 13 security updates introduce a new feature called Guarded Script. This mechanism restricts arbitrary code execution within the sandbox environment to prevent similar escape techniques in the future.
Compromised instances expose organizations to severe operational risks. Because the platform underpins critical enterprise workflows, attackers who achieve execution can establish persistence, move laterally across internal networks, and harvest credentials. Deploying autonomous tooling on these networks requires strict isolation protocols, similar to the ephemeral virtual machines outlined in modern agentic security guides.
If you manage on-premises or self-hosted ServiceNow deployments, you must apply the KB3137947 update immediately. You should also audit proxy server logs for unusual activity targeting the /assessment_thanks.do endpoint between April 1 and your patching date.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
How to Run TPU Workloads on Google Cloud with Ray 2.55
Learn how to provision Google Cloud TPUs, handle slice topologies, and deploy machine learning models using Ray 2.55 and the KubeRay Operator.
AutoJack Exploit Chains MCP Flaws for RCE in AutoGen
Microsoft patched AutoJack, a critical vulnerability chain in AutoGen Studio that enables remote code execution via AI browsing agents.
OpenClaw Users Warned to Assume Compromise After Major Breach
The popular OpenClaw AI agent framework faces a security crisis as researchers uncover critical RCE vulnerabilities and thousands of exposed instances.
Active RCE Exploits Target 7,000 Exposed Langflow Instances
Attackers are actively exploiting a path traversal vulnerability in Langflow's file upload endpoint to achieve unauthenticated remote code execution.
Hackers Exploit Critical RCE Flaw in Marimo Python Notebooks
A critical pre-auth vulnerability in Marimo is under active exploitation, allowing attackers to gain full shell access and steal sensitive API keys.