Keycloak EUVD-2026-92258

Exploitation requires that the target Keycloak realm has Dynamic Client Registration enabled and that the attacker holds a valid Initial Access Token authorizing client registration (reflected as PR:L). The specific flaw is that the registration process does not filter security-sensitive client attributes, allowing a registrant to set attributes that cause the new client to bypass audience (aud) validation during token introspection. Limiting factors: the attacker must already have been granted a registration token within the realm (not anonymous access), impact is confined to other applications in the same realm, and the impact is read-only confidentiality disclosure (identity data, roles, session details) with no integrity or availability effect. Realms without Dynamic Client Registration enabled, or that do not issue Initial Access Tokens to untrusted parties, are not exposed.

This is a moderate-priority authorization flaw (CWE-862, Missing Authorization) rather than a top-tier critical issue, and the signals are internally consistent. The CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N, base 6.5) indicates a network-reachable, low-complexity attack that nonetheless requires privileges (PR:L) - specifically a valid Initial Access Token for Dynamic Client Registration - and yields confidentiality-only impact (C:H, with I:N/A:N). No EPSS score, CISA KEV listing, SSVC decision data, or POC reference is present in the provided intelligence, so exploitation probability and automatability cannot be quantified from the input. Exploitation status is therefore 'no public exploit identified at time of analysis.' The main risk driver is that Keycloak is a widely deployed IAM platform and the flaw enables cross-application disclosure of identity data, roles, and session details within a shared realm; the main mitigant is the PR:L precondition (the attacker must already hold a valid Initial Access Token and the ability to register clients), which limits this to actors with some existing trust in the realm rather than anonymous internet attackers. Note that the input Tag 'Authentication Bypass' is imprecise - this is an authorization/audience-check bypass (consistent with CWE-862), not an authentication bypass; verify categorization with the Red Hat advisory.

An attacker who possesses a valid Initial Access Token uses Keycloak's Dynamic Client Registration endpoint to register a new client while injecting security-sensitive client attributes that the registration flow fails to filter. The malicious client is configured to bypass audience validation during token introspection, so the attacker can introspect access tokens issued to other applications in the same realm and read their identity claims, roles, and session details. No public exploit code is identified in the provided data, and the attack requires the pre-existing registration privilege, but exploitation complexity is low once that token is held.