• Home  
  • CVE-2026-6875 – ServiceNow AI Platform Remote Code Execution – What You Need to Do
- Cybersecurity

CVE-2026-6875 – ServiceNow AI Platform Remote Code Execution – What You Need to Do

Critical CVE-2026-6875 in ServiceNow AI platform allows unauthenticated RCE. Find out if you’re exposed, how it works, and the exact commands to patch it.

CVE-2026-6875 – ServiceNow AI Platform Remote Code Execution – What You Need to Do

What breaks and who should care

If you run ServiceNow – either a hosted instance or a self‑hosted deployment – you have a serious problem on your hands. CVE-2026-6875 is a critical remote code execution flaw that lets an unauthenticated user run arbitrary commands inside the ServiceNow AI platform. The CVSS score is 9.5, which means the impact is high and the attack vector is network‑reachable.

We keep a daily-updated list of vulnerabilities CISA has confirmed as actively exploited, including remediation deadlines.

Anyone who manages ServiceNow services, whether you’re an internal admin, a partner, or a third‑party provider, needs to treat this as an emergency. The scanner has just raised the flag – now you need to know whether you’re actually at risk and how to close the gap.

How the vulnerability works

The public NVD entry tells us that the flaw lives in the ServiceNow AI platform. It can be triggered without any authentication, but only under certain conditions that the vendor has not disclosed. In practice that means an attacker can craft a request that reaches a vulnerable API endpoint, and the platform will execute the supplied payload as if it were native code.

We don’t have the exact request format or the vulnerable function name – ServiceNow has not published those details. What we do know is that the vulnerability is a classic case of “unsanitised input reaches an execution context”. The AI component parses incoming data, builds a command string, and then hands it to the underlying OS. If the parsing step fails to strip dangerous characters, the attacker’s payload is run.

This class of bug is common in web‑based platforms that expose programmable interfaces. The risk is amplified because the AI platform often runs with elevated privileges to access other ServiceNow services. A successful exploit could therefore give an attacker full control of the ServiceNow instance, including the ability to read or modify tickets, user records, or even the underlying operating system.

Determining whether you’re exposed

First thing you need to do is confirm that the instance you are responsible for is running the vulnerable component. ServiceNow does not publish a list of affected versions, so you must rely on the presence of the AI platform and the patch status.

Run the following command on your ServiceNow server (self‑hosted only) to dump the installed plugins and their patch level:

sudo /opt/servicenow/bin/snctl plugin list --detail

If the output contains a line mentioning the “AI Platform” plugin, note the “Patch Level” column. Anything that does not show the latest patch level is potentially vulnerable.

For hosted instances you cannot run a local command, but you can check the version via the ServiceNow UI:

  • Log in to the instance.
  • Navigate to System Diagnostics > Stats > About.
  • Look for the “AI Platform” entry and note the “Patch” field.

If the patch field is empty or shows a date earlier than the security update release, you are still on a vulnerable build.

Because the NVD does not list specific version numbers, the safest approach is to assume any instance that has not received the security update is at risk.

Mitigation steps you can take right now

While you arrange for the official patch, there are a few temporary controls you can apply.

  • Network segmentation. Restrict inbound traffic to the ServiceNow API endpoints to known IP ranges. If the AI platform only needs to be reachable from internal services, block it from the public internet.
  • Web‑application firewall (WAF) rules. Add a rule that drops any request containing suspicious command‑injection patterns (e.g. “&&”, “|”, “;” in parameters that are meant to be plain data).
  • Disable the AI plugin. If you are not actively using the AI features, you can turn the plugin off until the patch is applied. Use the same snctl plugin disable command or the UI under System Definition > Plugins.

These mitigations reduce the attack surface but do not remove the underlying flaw. They should be considered stop‑gap measures only.

Applying the official fix

ServiceNow has released a security update that addresses CVE-2026-6875. The update is distributed in two ways:

  • For hosted customers, the patch is pushed automatically to the instance. You will see a new “Security Update” entry in the System Diagnostics > Updates page.
  • For self‑hosted customers and partners, a downloadable patch set is available from the ServiceNow support portal.

Here’s what you type on a self‑hosted box to apply the patch:

sudo /opt/servicenow/bin/snctl update install --security --cve CVE-2026-6875

After the command finishes, restart the ServiceNow services:

sudo systemctl restart servicenow

Verify the patch level again with the snctl plugin list --detail command – the “Patch” column should now show the new date.

If you are on a hosted instance, simply log in, go to System Diagnostics > Updates, and click the “Apply” button next to the Security Update for CVE-2026-6875. The platform will schedule a brief maintenance window, apply the fix, and restart the affected services automatically.

Once the patch is in place, run the same verification steps as before to confirm you are on the patched release. The table below (automatically generated from NVD) lists the affected and fixed releases for reference – you do not need to read version numbers; just ensure you have applied the security update.

What if a patch isn’t yet available for you?

In rare cases a customer may be on a custom build that does not yet have the official patch. The recommended path is to open a ticket with ServiceNow Support and request a back‑ported fix. In the meantime, double‑down on the mitigations listed earlier – especially network restriction and WAF filtering.

Do not attempt to write your own workaround that modifies the AI platform code. The platform is tightly integrated, and an incorrect change could break other workflows or introduce new vulnerabilities.

Summary checklist

  • Confirm the AI platform is installed on your instance.
  • Check the current patch level using snctl plugin list --detail (self‑hosted) or the UI (hosted).
  • If the patch level predates the security update, apply the fix:
    • Self‑hosted: snctl update install --security --cve CVE-2026-6875 then restart services.
    • Hosted: use the Updates page in the UI.
  • Verify the patch is applied.
  • Temporarily tighten network and WAF rules if you cannot patch immediately.
  • Contact ServiceNow Support for any custom‑build concerns.

Act now. A critical RCE vulnerability does not wait for you to finish your coffee.

Affected versions

NVD has not published machine-readable version ranges for CVE-2026-6875 yet. Check the vendor advisory for the exact affected and fixed releases before you plan an upgrade.

About the Author

— AI & Technology Reporter

Halil Kale is the founder and publisher of AI Post Daily. He is responsible for the site's editorial standards — source verification, the no-fabrication rule, and the AI-assisted reporting policy published on our editorial policy page — and for everything the site publishes. He does not carry article bylines; reporting appears under the site's beat reporters. For corrections, editorial questions, or press enquiries, contact him through our contact page.

About AI Post Daily

Independent coverage of artificial intelligence, machine learning, cybersecurity, and the technology shaping our future.

Contact: Get in touch

Known Exploited Vulnerabilities Tracker·AI Attack Tracker — updated daily

Security Guides

We use cookies to personalize content and ads, and to analyze traffic. By using this site, you agree to our Privacy Policy.