Critical ServiceNow AI Platform Flaw Exploited for Unauthenticated Code Execution

Unauthenticated Code Execution on ServiceNow

ServiceNow, a prominent provider of digital workflow solutions, is grappling with a critical security vulnerability, CVE-2026-6875, affecting its AI Platform. This flaw, rated with a severe CVSS score of 9.5, is categorized as a sandbox escape vulnerability. Its severity stems from the fact that it allows an unauthenticated user to execute arbitrary code on affected instances. This means an attacker doesn't need legitimate credentials to gain control, significantly lowering the bar for exploitation and increasing the potential impact on organizations relying on ServiceNow for their operations.

The vulnerability was initially reported by Searchlight Cyber on April 1, 2026, highlighting its potential for a complete compromise of the ServiceNow instance, extending even to connected proxy servers. Such a broad impact underscores the architectural significance of the flaw, suggesting it could bypass multiple layers of security designed to isolate and protect core systems. The disclosure of technical specifics by Searchlight Cyber likely provided threat actors with the necessary information to develop and deploy exploits, accelerating the timeline from disclosure to active exploitation.

The Active Exploitation Campaign

Threat intelligence firm Defused Cyber has confirmed active exploitation of CVE-2026-6875 in the wild. Attackers are targeting a specific pre-authentication endpoint, "/assessment_thanks.do," using HTTP POST requests. While the exact sandbox-escape gadget used in these attacks may differ from publicly documented proof-of-concept (PoC) exploits, the ultimate outcome remains the same: arbitrary code execution. This indicates that threat actors are adapting their methods or discovering alternative routes to achieve the same critical primitive, demonstrating a sophisticated understanding of the vulnerability and the underlying platform.

The observed exploitation efforts underscore the urgency for organizations to apply patches. The fact that attackers are actively leveraging this flaw means that any unpatched ServiceNow instance is a direct target, facing imminent risk of compromise. The use of a pre-authentication endpoint is particularly concerning, as it allows attackers to initiate their malicious activities without needing to bypass any initial authentication mechanisms, making the attack surface much larger and easier to target.

Mitigating the Critical Threat

ServiceNow has responded to the vulnerability by releasing patches throughout June 2026 across various versions, including Brazil EA and GA, Australia Patch 2, Zurich Patch 7b and 9, and Yokohama Patch 12 Hot Fix 1b and Patch 13. These updates are crucial for mitigating the risk posed by CVE-2026-6875. Beyond just patching, ServiceNow is also enhancing instance security by severely restricting the type of code that can run in sandbox contexts, a proactive measure aimed at preventing similar sandbox escape vulnerabilities in the future.

For customers running self-hosted versions of ServiceNow, the immediate application of these fixes is paramount. Organizations must prioritize patching efforts to protect their instances from ongoing exploitation. Failure to do so could lead to unauthorized access, data exfiltration, system disruption, and significant reputational damage. The combination of active exploitation and the critical nature of the flaw makes this a top-tier security concern requiring swift and decisive action from all affected ServiceNow users.