Elastic Kibana EUVD-2026-116073

Requires an authenticated Kibana user who holds Machine Learning read access in at least one space, and a multi-space deployment where data frame analytics jobs exist in other spaces the user is not authorized to access. Disclosure is limited to the audit and error messages of those data frame analytics jobs (destination index names, analyzed field names, failure information) - not to the underlying analyzed data, and with no integrity or availability impact. Limiting factors: valid credentials with ML read privileges are mandatory (PR:L, so not exploitable by anonymous/unauthenticated attackers), and the issue is only meaningful in deployments that use multiple Kibana spaces with separated ML workloads. Single-space deployments or users with no ML access are not affected.

This is a low-to-moderate priority issue, not an emergency. The CVSS 3.1 base score of 4.3 (AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N) reflects a network-reachable but authenticated flaw: the attacker must already hold Machine Learning read access in at least one Kibana space, and the only impact is low-confidentiality information disclosure with no integrity or availability effect. The vendor's priority label is 'this_month' (not act_now or emergency), which is consistent with the modest technical impact. No EPSS score was provided in the input, so exploitation probability cannot be quantified. The CVE is not listed in CISA KEV, so there is no public evidence of active exploitation. No proof-of-concept or exploit reference is present - the only reference is the Elastic security advisory (ESA-2026-203), not exploit code. The signals are internally consistent: a genuine but limited cross-space authorization leak affecting only multi-space deployments with multiple ML-authorized users, disclosing metadata rather than record-level data.

An authenticated Kibana user who has been granted Machine Learning read access within one space queries the data frame analytics job audit and error messages and, due to a user-controlled key authorization check (CWE-639), receives messages belonging to jobs in other spaces they should not see. This leaks job metadata such as destination index names, analyzed field names, and failure details, aiding reconnaissance of other teams' data pipelines. No public exploit code is identified at time of analysis; the attack requires valid ML read credentials and offers only low-confidentiality disclosure.