Critical ServiceNow AI Platform Vulnerability Comes Under Active Attack

Threat intelligence firm Defused has reported that a critical vulnerability, CVE-2026-6875, in the ServiceNow AI Platform is now being exploited. Researchers observed the first attack attempts on Friday — just days after the vendor shipped its fixes.

The ServiceNow AI Platform (formerly known as the Now Platform) is an enterprise PaaS offering that helps businesses embed artificial intelligence into core workflows. According to the company itself, the platform handles more than 100 billion workflows a year and powers over 100,000 enterprise AI applications across 85% of the Fortune 500. Its scale is what makes the vulnerability especially dangerous.

What the vulnerability allows

The flaw was discovered and reported on April 1 by Searchlight Cyber. According to its description, the vulnerability lets an unauthenticated attacker break out of the sandbox and remotely execute arbitrary code inside the platform. That said, the attack is rated as high-complexity.

ServiceNow addressed the issue on its self-hosted instances and, on July 13 — a week ago — also released security updates for self-managed deployments.

Attacks in the wild

On Saturday, Defused warned of active exploitation of a “pre-authentication sandbox-escape RCE.” The researchers noted a curious detail: the payload lands at the same point described by Searchlight Cyber (the /assessment_thanks.do handler), yet the sandbox-escape mechanism reaches the same result by a different route than the published PoC. In other words, the attackers didn't simply replay a ready-made example — they found their own path to code execution.

Interestingly, ServiceNow itself has yet to acknowledge the exploitation: its official advisory still states that the company “is not aware of any attacks on ServiceNow instances.” Even so, the vendor strongly urges any customers who have not yet applied the update to do so as soon as possible. When BleepingComputer asked the company to confirm Defused's findings, a spokesperson was unable to respond promptly.

This is not the platform's first security incident of late. A month earlier, ServiceNow privately disclosed a case in which attackers could query data from customer instances without authentication through a vulnerable API endpoint. That episode, however, the company attributed not to malicious actors but to researchers and customers operating within its bug bounty program.

Source: Bleeping Computer