CVE-2026-6875: ServiceNow AI Platform RCE Hits Fortune 500 Scale

A pre-authentication remote code execution vulnerability in ServiceNow’s AI Platform is being actively exploited in the wild, with attack traffic first confirmed by threat intelligence firm Defused in late July 2026. The flaw, tracked as CVE-2026-6875 with a CVSS score of 9.5, lets any unauthenticated attacker with network access to a vulnerable instance escape ServiceNow’s server-side scripting sandbox and run arbitrary code on the underlying host.

The numbers behind this make it uncomfortable reading: ServiceNow’s AI Platform processes more than 100 billion enterprise workflows annually and is deployed at around 85% of Fortune 500 companies. This is as close to ubiquitous enterprise software as it gets.

How It Works

Searchlight Cyber researchers discovered and reported the vulnerability to ServiceNow in early April 2026. The flaw lives in ServiceNow’s Rhino JavaScript sandbox, the environment that executes server-side scripts in ServiceNow workflows and integrations. A code injection in how the sandbox processes certain inputs allows an attacker to break out of the sandboxed context entirely.

The vulnerable entry point is the unauthenticated endpoint /assessment_thanks.do. An attacker hitting this endpoint with a crafted payload can trigger sandbox escape and reach unsandboxed code execution on the ServiceNow host. From there, the attacker can read and write any data table in the instance, create administrator accounts, and send commands to ServiceNow’s MID Server proxies.

MID Servers are the piece that makes this especially significant. They are ServiceNow’s agents for internal network integration, and they typically run with broad permissions inside the corporate network. Compromising a MID Server turns the ServiceNow instance into a pivot into the internal network — a bridge from the exploit to everything that ServiceNow integrates with: Active Directory, CMDB, ITSM tickets, HR data, security tooling.

ServiceNow patched hosted (cloud) instances on the same day it received the report and made patches available to self-hosted customers through June 2026. The vulnerability was publicly disclosed on July 13. Active exploitation was confirmed within days of public disclosure.

Why “Already Patched” Isn’t the End of the Story

The patch timeline sounds clean. In practice, self-hosted ServiceNow deployments on enterprise schedules can lag behind, and hybrid deployments where some components are self-managed complicate the picture further. The exploit is being used against a different gadget chain than the one Searchlight Cyber publicly documented — meaning attackers have independent working exploits that don’t rely on proof-of-concept code.

Defused reported hitting the same pre-authentication entry point documented publicly, but via a distinct exploitation path. That matters because it suggests multiple threat actors have developed working exploits independently, not simply copied from a single public PoC. Expect continued exploitation across a variety of attacker profiles.

What’s Actually at Risk

The standard enterprise ServiceNow deployment is deeply integrated. ServiceNow touches HR workflows, IT service management, procurement approvals, change management, and increasingly AI-driven process automation. An attacker with full instance access sees all of that.

More specifically, and more immediately dangerous: ServiceNow instances often hold active credentials and API tokens for the services they integrate with. A compromised instance is frequently the fastest path to credentials for adjacent systems, well before an attacker needs to do anything sophisticated in the internal network.

For organisations running ServiceNow with AI features enabled — Predictive Intelligence, Now Assist, Virtual Agent, and similar — the instance also processes data that feeds AI models. Access to that data and the ability to inject into AI workflows adds a dimension that goes beyond conventional ITSM breach impact.

Checking Your Exposure

The fastest check: test whether /assessment_thanks.do on your ServiceNow instance returns a response without authentication. If it does and you haven’t applied the patch, you’re exposed.

For fully cloud-hosted ServiceNow customers, ServiceNow applied the patch automatically in April. Verify your instance version in System Diagnostics and confirm the patch level matches ServiceNow’s published security advisory.

For self-hosted and hybrid deployments, the patch installation requires a maintenance window. If that window hasn’t happened yet, contact your ServiceNow implementation partner today, not next week.

References

• Help Net Security — ServiceNow pre-auth RCE exploited in the wild (CVE-2026-6875)
• The Hacker News — Critical ServiceNow AI Platform Flaw Exploited for Unauthenticated Code Execution
• BleepingComputer — Critical ServiceNow code execution flaw now exploited in attacks
• Security Affairs — Attackers Exploit Critical ServiceNow RCE Flaw CVE-2026-6875
• Indusface — CVE-2026-6875: ServiceNow Sandbox Escape to Pre-Auth RCE