Three of the four flaws ServiceNow patched in late August carry a score of 10.0, the top of the CVSS scale, and none of them needs a login to exploit. The Hacker News reported the fixes on August 28, a day after ServiceNow published its advisory for the AI Platform.
ServiceNow said it has already pushed a security update to the instances it hosts and handed the same update to partners and self-hosted customers. That last group is the one with work to do. Anyone running their own instance has to apply the patches.
The four flaws
CVE-2026-18885 is a code injection bug in the GraphQL Composite Data API. In the right circumstances an unauthenticated user could run arbitrary code and read or change instance data.
CVE-2026-18886 sits in the system configuration image upload processor. It is an improper access control problem that could let an unauthenticated user create or alter instance data and escalate privileges.
CVE-2026-74820 is SQL injection, reached through a dynamic schema ORDER BY clause. An unauthenticated attacker could run arbitrary SQL against the database underneath the instance.
The fourth, CVE-2026-6876, scores 8.7. It is a sandbox escape in the Now Platform. ServiceNow describes it as letting an unauthenticated user execute code, yet the vector it assigned says low privileges are required, and it records no impact beyond the vulnerable component itself.
The three top-rated flaws share one vector: reachable over the network, low complexity, no privileges, no user interaction, and high impact on confidentiality, integrity and availability, both in the affected component and in connected systems.
A rating nobody else has checked
The 10.0 scores are ServiceNow’s own. It is the CVE Numbering Authority for its products. And since April 15, 2026, NIST has enriched only vulnerabilities that sit in CISA’s Known Exploited Vulnerabilities catalog, touch federal government software, or are designated critical under Executive Order 14028. None of the four was in the catalog as of August 28, so the vendor’s number is the only severity assessment on record.
ServiceNow said in each of the four records that it is not currently aware of exploitation. The Hacker News found no public exploit code for the three maximum-severity flaws, and Searchlight Cyber had published no technical write-up of the August bugs when the story ran.
Affected versions span several release families. The advisory covers Xanadu releases before Patch 11 Hot Fix 7a, Yokohama releases before Patch 12 Hot Fix 3b or Patch 13 Hot Fix 4, and Zurich and Australia releases below a list of named patches, ending at Zurich Patch 12 and Australia Patch 5. One oddity: the record for CVE-2026-18886 lists Australia releases before Patch 5 as “unknown”, where the other three records mark them affected. All four records default to unaffected, so a release the advisory does not name falls outside the affected set.
The July precedent
The advisory follows CVE-2026-6875, a pre-authentication sandbox escape in the same platform. Searchlight Cyber reported it to ServiceNow on April 1, 2026, and ServiceNow published its advisory on July 13. Days later, the threat intelligence firm Defused said it was seeing the flaw exploited in the wild. It then corrected itself: the payload it captured matched Searchlight Cyber’s published proof-of-concept exploit.
ServiceNow scored that bug 9.5 under the same version of the scoring system. Every metric matched the new trio except attack complexity, which the company set to high.
A ServiceNow spokesperson told The Hacker News that “based on our investigation to date, we have not observed evidence that this activity is related to instances that ServiceNow hosts.” The company, the spokesperson added, encourages self-hosted and hosted customers alike to apply the patches and will keep working directly with customers who need help. In July, Searchlight Cyber researcher Adam Kues wrote that ServiceNow was “enhancing instance security by severely restricting the type of code that can run in sandbox contexts.”
