Critical Flaws in ServiceNow AI Platform: Three Max-Severity Vulnerabilities Patched
ServiceNow has issued critical patches for four security vulnerabilities affecting its AI Platform, three of which carry a maximum CVSS score of 10.0. These flaws, if exploited under specific conditions, could allow unauthenticated attackers to execute arbitrary code, gain unauthorized access, or modify sensitive instance data. Organizations running self-hosted instances are urged to apply these fixes immediately.

**ServiceNow** has recently addressed four significant security vulnerabilities within its **AI Platform**, with three of these flaws receiving a critical **CVSS** score of 10.0. These high-severity issues could be exploited by unauthenticated attackers under certain conditions.
### Patches Deployed and Urgency for Self-Hosted Instances
On August 27, 2026, **ServiceNow** rolled out security updates to its hosted instances and distributed the patches to partners and customers operating self-hosted environments. This places the onus on organizations managing their own instances to promptly apply the necessary fixes.
### The Critical Vulnerabilities
Below are the details of the four vulnerabilities, with three standing out due to their perfect CVSS score:
* **CVE-2026-18885** (CVSS score: 10.0): A code injection flaw in the **GraphQL Composite Data API** that could allow an unauthenticated user to execute arbitrary code and gain access to or modify instance data.
* **CVE-2026-18886** (CVSS score: 10.0): An improper access control vulnerability in the system configuration image upload processor, potentially enabling an unauthenticated user to create or modify instance data, leading to privilege escalation.
* **CVE-2026-74820** (CVSS score: 10.0): A **SQL injection** vulnerability accessible via a dynamic schema `ORDER BY` clause, which could allow an unauthenticated user to execute arbitrary SQL statements against the underlying database.
* **CVE-2026-6876** (CVSS score: 8.7): A sandbox escape within the **Now Platform** that could permit an unauthenticated user to execute arbitrary code.
All three maximum-severity flaws share a common attack vector: `CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H`. This vector describes a network-reachable attack of low complexity that requires no privileges or user interaction, leading to a high impact on confidentiality, integrity, and availability for both the vulnerable component and connected systems.
### Precedent: CVE-2026-6875 and In-the-Wild Exploitation
These new advisories follow **CVE-2026-6875**, another pre-authentication sandbox escape in the same platform. **Searchlight Cyber** reported this flaw to **ServiceNow** on April 1, 2026, with an advisory published on July 13.
Shortly after the July advisory, threat intelligence firm **Defused** reported observing in-the-wild exploitation of **CVE-2026-6875**. A subsequent correction clarified that the captured payload matched **Searchlight Cyber's** [published proof-of-concept (PoC) exploit](https://www.slcyber.io/research/smashing-the-servicenow-sandbox-pre-authentication-rce).
A **ServiceNow** spokesperson addressed the matter, stating, "Based on our investigation to date, we have not observed evidence that this activity is related to instances that **ServiceNow** hosts." They further urged all customers to apply relevant patches.
### ServiceNow's Role as CVE Numbering Authority
**ServiceNow** acts as the **CVE Numbering Authority** for its products. Since April 15, 2026, **NIST** has primarily focused on enriching vulnerabilities that appear in **CISA's Known Exploited Vulnerabilities catalog**, affect federal government software, or are deemed critical under Executive Order 14028. As of August 28, 2026, none of the four newly disclosed flaws were in **CISA's** catalog, making **ServiceNow's** ratings the sole severity assessment available.
Interestingly, **ServiceNow** rated the sandbox escape reported exploited in July (**CVE-2026-6875**) at 9.5, with an identical metric set to the current 10.0 flaws, except for attack complexity, which was set to high. The newly disclosed maximum-severity flaws are all rated with low attack complexity.
### Affected Versions
The August advisory from **ServiceNow** lists the following versions as affected:
* **Xanadu**: Any version before Patch 11 Hot Fix 7a
* **Yokohama**: Any version before Patch 12 Hot Fix 3b, and any version before Patch 13 Hot Fix 4
* **Zurich**: Any version before Patch 7b Hot Fix 3, Patch 8 Hot Fix 5, Patch 9 Hot Fix 6, Patch 10 Hot Fix 2m (m-branch), Patch 10 Hot Fix 3 (standard), Patch 11, or Patch 12
* **Australia**: Any version before Patch 2 Hot Fix 3, Patch 3 Hot Fix 2, Patch 3m, Patch 4, or Patch 5
It's noted that the record for **CVE-2026-18886** marks "Any version before Australia Patch 5" with an unknown status, while the other three mark it as affected. All four advisories default to an unaffected product status for releases not explicitly named.
### Discrepancies and Current Exploitation Status
While **ServiceNow** describes **CVE-2026-6876** as allowing an unauthenticated user to execute arbitrary code, its assigned CVSS vector specifies `PR:L` (low privileges required). This vector also indicates no impact beyond the vulnerable component, unlike the three 10.0-rated flaws.
**ServiceNow** has stated that it is not currently aware of exploitation for any of the four newly disclosed vulnerabilities. As of August 28, 2026, no public exploit code was found for the three maximum-severity flaws. **Searchlight Cyber** had not published a technical write-up for these August-disclosed flaws at the time of writing, though **Adam Kues**, a security researcher at the firm, noted in July that **ServiceNow** was "enhancing instance security by severely restricting the type of code that can run in sandbox contexts."