Attackers have begun exploiting a critical vulnerability in the ServiceNow AI Platform (formerly the Now Platform), according to threat intelligence company Defused. The flaw, tracked as CVE-2026-6875, allows unauthenticated threat actors to escape the platform’s sandbox and achieve remote code execution, though the attacks are described as high-complexity.
The vulnerability was discovered and reported by cybersecurity firm Searchlight Cyber on April 1st. ServiceNow patched hosted instances and released updates for self-hosted deployments on July 13th, roughly a week before exploitation was observed in the wild.
Defused researchers said the first exploitation attempts appeared on Friday. In a Saturday post, the company stated it was observing in-the-wild exploitation of the pre-auth sandbox-escape RCE, noting that the payloads hit the same pre-auth sink documented by Searchlight Cyber, the assessment_thanks.do endpoint, but reach code execution through a different route than the originally published proof-of-concept.
Vendor Response
As of publication, ServiceNow’s official advisory still states the company is not aware of exploitation against its instances, despite Defused’s findings. ServiceNow continues to urge all customers who haven’t applied the patch to upgrade immediately. A ServiceNow spokesperson did not immediately respond to a request for comment confirming active exploitation.
Second Incident Disclosed
This is the second ServiceNow security issue to surface recently. Last month, the company privately disclosed an incident in which attackers queried data from customer instances by exploiting an unauthenticated access flaw in a vulnerable API endpoint. ServiceNow later attributed that activity to security researchers or bug-bounty-linked research rather than malicious actors.
Why It Matters
ServiceNow’s AI Platform is deeply embedded in enterprise operations, running more than 100 billion workflows annually and powering over 100,000 enterprise AI applications at 85% of Fortune 500 companies. A pre-authentication RCE flaw at this scale represents significant risk for organizations that have not yet applied the July 13th patch.
Security teams running self-hosted ServiceNow instances should prioritize patching immediately and review logs for requests to the assessment_thanks.do endpoint as a potential indicator of exploitation attempts.