Skip to main content
CVE Vulnerability Database
Vulnerability Database/CVE-2026-78606

CVE-2026-78606: Elastic Kibana Auth Bypass Vulnerability

CVE-2026-78606 is an authorization bypass flaw in Elastic Kibana affecting users with shared usernames across authentication realms. This article covers technical details, affected versions, and mitigation.

Published:

CVE-2026-78606 Overview

CVE-2026-78606 is an Incorrect Authorization vulnerability [CWE-863] in Elastic Kibana. The flaw affects the Elastic AI Assistant Knowledge Base component. When two authenticated principals from different authentication realms share the same username value, one principal can read, modify, and delete the other's private Knowledge Base entries. The issue stems from Kibana treating the username string as a unique identifier without accounting for the originating realm. Elastic addressed the vulnerability in Kibana versions 8.19.21, 9.4.6, and 9.5.2 under advisory ESA-2026-142.

Critical Impact

Authenticated users sharing a username across authentication realms can access, alter, or delete another user's private Elastic AI Assistant Knowledge Base entries.

Affected Products

  • Elastic Kibana versions prior to 8.19.21
  • Elastic Kibana 9.x versions prior to 9.4.6
  • Elastic Kibana 9.5.x versions prior to 9.5.2

Discovery Timeline

  • 2026-09-01 - CVE-2026-78606 published to NVD
  • 2026-09-02 - Last updated in NVD database

Technical Details for CVE-2026-78606

Vulnerability Analysis

The vulnerability resides in the authorization logic that scopes private Elastic AI Assistant Knowledge Base entries to their owning user. Kibana identifies entry ownership using only the username attribute of the authenticated principal. When Kibana is configured with multiple authentication realms, such as native, LDAP, SAML, or Active Directory, two distinct principals can legitimately share the same username value.

Because the authorization check does not incorporate the realm identifier, Kibana treats these distinct users as the same owner. A user authenticating from one realm can therefore enumerate, read, modify, or delete Knowledge Base entries created by a different user with the same username in another realm. The impact is limited to the Knowledge Base data surface of the Elastic AI Assistant.

Root Cause

The root cause is an incomplete authorization key. Ownership records rely on username alone rather than a composite identifier that combines both username and the authentication realm name. This design assumption fails as soon as the deployment includes two or more realms with overlapping username namespaces.

Attack Vector

Exploitation requires an authenticated Kibana account and knowledge of a target username that exists in a separate realm. The attacker authenticates through their own realm, then issues Elastic AI Assistant Knowledge Base API calls that resolve to the victim's entries. No user interaction from the victim is required. The attack complexity is elevated because the attacker must operate in a deployment configured with multiple realms and must know or guess a colliding username.

See the Elastic Security Update ESA-2026-142 for vendor technical details.

Detection Methods for CVE-2026-78606

Indicators of Compromise

  • Elastic AI Assistant Knowledge Base entries that appear modified or deleted without a corresponding action from the owning user.
  • Audit log entries showing Knowledge Base API calls from a user whose realm differs from the entry's original creator realm.
  • Unexpected read or write activity on /internal/elastic_assistant/knowledge_base/entries endpoints from accounts that do not typically use the AI Assistant.

Detection Strategies

  • Enable Kibana audit logging and correlate Knowledge Base API events with the authentication_realm field of the acting principal.
  • Compare the realm of the current requester against the realm recorded when each Knowledge Base entry was created and flag mismatches.
  • Inventory usernames across all configured authentication realms and alert on collisions that could enable cross-realm access.

Monitoring Recommendations

  • Ship Kibana and Elasticsearch audit logs to a centralized analytics platform and retain them for post-incident review.
  • Alert on bulk read, modify, or delete operations against Knowledge Base entries within short time windows.
  • Review authentication realm configuration changes and new realm additions for governance approval.

How to Mitigate CVE-2026-78606

Immediate Actions Required

  • Upgrade Kibana to version 8.19.21, 9.4.6, or 9.5.2 or later as documented in advisory ESA-2026-142.
  • Audit configured authentication realms and identify any usernames that collide across realms.
  • Review existing Elastic AI Assistant Knowledge Base entries for unauthorized modification or deletion.

Patch Information

Elastic released fixed versions 8.19.21, 9.4.6, and 9.5.2 under Elastic Security Update ESA-2026-142. The patch updates the ownership check to include the authentication realm alongside the username, preventing cross-realm access to private Knowledge Base entries.

Workarounds

  • Enforce unique username namespaces across all configured authentication realms until the upgrade can be applied.
  • Restrict access to the Elastic AI Assistant Knowledge Base feature to a limited set of trusted users while the patch is pending.
  • Temporarily disable additional authentication realms if their user populations cannot be deconflicted from the primary realm.
bash
# Verify installed Kibana version and confirm it meets the patched release
curl -s -u "$KBN_USER:$KBN_PASS" "$KBN_URL/api/status" | jq '.version.number'

# List configured authentication realms in Elasticsearch
curl -s -u "$ES_USER:$ES_PASS" "$ES_URL/_security/_authenticate" | jq '.authentication_realm'

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Default Legacy - Prefooter | Experience the World’s Most Advanced Cybersecurity Platform

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.