ServiceNow AI Platform EUVD-2026-86395

Exploitation targets the ServiceNow AI Platform and, per the CVSS vector, requires no authentication, no user interaction, and only network reachability to the instance (AV:N/PR:N/UI:N/AC:L). However, ServiceNow explicitly limits this to 'certain circumstances,' meaning it is NOT universally exploitable against every default deployment - undisclosed configuration or state preconditions gate the flaw, though the vendor did not name them. Practical limiting factors: hosted (SaaS) instances were already patched by ServiceNow in August 2026, so residual exposure is concentrated in self-hosted and partner-managed instances that have not yet applied the update. Impact is confined to data integrity (create/modify/delete records); there is no confidentiality or availability impact per the vector.

The vendor-supplied CVSS 4.0 vector (CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N, score 8.7) describes a network-reachable, low-complexity, pre-authentication (PR:N) improper-access-control flaw whose sole impact is integrity (VI:H) - the ability to create, modify, or delete instance data - with no confidentiality (VC:N) or availability (VA:N) impact. There is no EPSS score, no CISA KEV listing, and no public exploit in the provided data; ServiceNow explicitly states it is 'not currently aware of malicious exploitation.' So this is a genuinely high-severity, unauthenticated data-tampering issue on a widely deployed enterprise platform, but with no evidence of active exploitation to date, priority is driven by the severity and pre-auth reachability rather than by an in-the-wild signal. Two caveats must be stated explicitly: (1) the description says the flaw is exploitable only 'in certain circumstances,' which is in tension with the AC:L/PR:N vector - the true attack complexity and required preconditions are not disclosed, so real-world exploitability may be narrower than the score implies; (2) the input Tag 'Information Disclosure' contradicts the CVSS vector, which shows VC:N and VI:H - this is a data-integrity/tampering issue, not a disclosure issue.

An unauthenticated remote attacker sends crafted requests to an exposed ServiceNow AI Platform instance and, under the conditions the vendor left unspecified, bypasses access controls to create, alter, or delete records beyond their intended scope - for example tampering with or destroying business records. No public proof-of-concept is available and the vendor reports no known exploitation, and the attack is limited by the 'in certain circumstances' preconditions noted in the advisory.