ServiceNow AI Platform Flaw (CVE-2026-6875) Now Under Active Exploitation

What CVE-2026-6875 actually is

CVE-2026-6875 is a pre-authentication sandbox-escape vulnerability in ServiceNow's AI Platform (formerly the Now Platform), rated 9.5 on CVSS. Security researchers at Searchlight Cyber found that attacker-controlled input reaching a server-side GlideRecord query path — via the unauthenticated /assessment_thanks.do endpoint — could trigger JavaScript evaluation that escapes ServiceNow's script sandbox. In practice, that means an attacker with no credentials and no prior foothold can potentially read sensitive platform data, create administrative accounts, and execute code, including through configured MID Servers or proxy infrastructure.

Timeline: from private report to public exploitation

• April 1, 2026 — Searchlight Cyber reports the flaw to ServiceNow.
• April 2, 2026 — ServiceNow ships a cloud-side mitigation for its own hosted instances, one day after the report.
• June 2026 — Patches for self-hosted customers and partners roll out across affected release trains.
• July 13, 2026 — Technical details are publicly disclosed.
• Mid-to-late July 2026 — Threat-intel firm Defused reports observing active exploitation in the wild.

Why this exploitation wave is notable

Defused's telemetry shows attackers hitting the same /assessment_thanks.do sink Searchlight Cyber documented, but reaching code execution through a different sandbox-escape gadget than the one in the original proof-of-concept. That matters operationally: detections or WAF signatures tuned narrowly to the published PoC may miss this variant. Searchlight Cyber researcher Adam Kues has described the worst-case outcome as complete compromise of the ServiceNow instance and any connected proxy servers — not just data exposure.

Who's actually exposed

Risk here is concentrated in self-hosted and on-prem ServiceNow deployments that haven't applied the June 2026 updates — specifically unpatched Brazil (EA/GA), Australia Patch 2, Zurich (Patch 7b/9), and Yokohama (Patch 12 Hot Fix 1b/13) releases. ServiceNow says it has "not observed evidence that this activity is related to instances that ServiceNow hosts," since those were covered by the April cloud-side fix, and is urging customers who manage their own instances to patch immediately if they haven't already.

What to do now

• Confirm whether your ServiceNow deployment is self-hosted/on-prem, and if so, apply the June 2026 security updates without delay.
• Check ServiceNow access and application logs for unexpected hits to /assessment_thanks.do, particularly from unauthenticated sources.
• Review MID Server and outbound proxy configurations — Searchlight Cyber flagged these as a path to broader compromise once the sandbox is escaped.
• If your instance was internet-facing and unpatched at any point since the July 13 disclosure, treat it as a candidate for a forensic review rather than assuming the patch alone resolves prior exposure.

FAQ