CVE-2026-6875 — a critical pre-auth sandbox-escape remote code execution flaw in the ServiceNow AI Platform — is being exploited in the wild, security researchers say.
CVE-2026-6875: what the flaw does
Searchlight Cyber, which discovered and reported the vulnerability on April 1, said the flaw “allows unauthenticated threat actors to escape the sandbox and execute code remotely within the ServiceNow platform in high-complexity attacks.” The vulnerability affects the ServiceNow AI Platform (formerly the Now Platform), an enterprise PaaS used to integrate AI into core workflows.
ServiceNow itself highlights the platform’s reach: it “runs more than 100 billion workflows each year and powers over 100,000 enterprise AI apps at 85% of all Fortune 500 companies.” That scale is the reason the vulnerability’s potential impact drew rapid attention from both defenders and researchers.
Discovery, patching, and the timeline to exploitation
Searchlight Cyber publicly disclosed the issue on April 1. ServiceNow addressed the flaw across hosted instances and published security updates for self-hosted instances on July 13. Defused, a threat intelligence company, reported over the weekend that attackers had begun exploiting CVE-2026-6875 in the wild, with “the first attempts being observed on Friday,” only days after ServiceNow issued its self-hosted patches.
What Defused observed and how attackers reached code execution
Defused researchers warned on Twitter: “We are observing in-the-wild exploitation of the ServiceNow pre-auth sandbox-escape RCE (CVE-2026-6875).” They said the active payloads “hit the same pre-auth sink @SLCyberSec documented (/assessment_thanks.do), but the sandbox-escape gadget reaches the same code-execution primitive by a different route than their published PoC.”
In short: the exploit traffic taps a previously documented pre-auth endpoint, but uses a different sandbox-escape gadget to convert that access into remote code execution, according to Defused’s post.
ServiceNow’s public posture and its recent disclosure history
ServiceNow’s official advisory — as of the reporting in BleepingComputer — still states it is “not currently aware of exploitation against ServiceNow instances.” The company advises customers who have not yet patched to “secure their systems against attacks by upgrading to a patched release as soon as possible.”
A ServiceNow spokesperson was “not immediately available for comment,” BleepingComputer reported when it reached out to confirm Defused’s finding that CVE-2026-6875 is now actively exploited.
Separately, ServiceNow privately disclosed a security incident last month in which attackers queried data from customer instances by exploiting an unauthenticated access flaw via a vulnerable API endpoint. In a subsequent advisory, ServiceNow tied that incident to “security researchers or customer-led research linked to bug bounty submissions” rather than to malicious threat actors.
What this means for technologists, enterprise customers, and threat actors
- Technologists and security teams: ServiceNow’s July 13 security updates for self-hosted instances are the immediate mitigation. Teams handling hosted instances should confirm ServiceNow’s remediation status, and those responsible for self-hosted deployments must apply the CVE-2026-6875 updates as soon as possible.
- Affected enterprises and procurement leaders: Because ServiceNow’s AI Platform is widely embedded — “over 100,000 enterprise AI apps” and use across “85% of all Fortune 500 companies” — procurement and incident-response leaders will need to verify patch deployment and assess any unusual activity related to the documented pre-auth sink (/assessment_thanks.do).
- Adversaries and threat actors: Defused’s confirmation that attackers have already begun targeting the vulnerability indicates exploitation techniques are circulating and that multiple routes to the same code-execution primitive exist, increasing urgency for defenders to apply patches.
The basic tension in this episode is plain in the record: a public researcher disclosed a critical sandbox-escape RCE in April, ServiceNow released hosted mitigations and a July 13 update for self-hosted instances, and within days threat intelligence observed active exploitation using a variant route to the same code-execution primitive. ServiceNow continues to advise patching while its public advisory maintains it is “not currently aware of exploitation,” and BleepingComputer’s outreach for comment found the company’s spokesperson “not immediately available for comment.”
For now the clearest, documented next step is technical: apply the CVE-2026-6875 updates to self-hosted instances and verify hosted-instance mitigations. Beyond that, the record leaves open the scope of successful intrusions tied to the weekend activity and whether ServiceNow will update its public advisory in response to Defused’s observations.
