Kyverno versions prior to 1.19.2 allow an authorization bypass through the Generation Handler by spoofing internal generate.kyverno.io labels. This flaw enables an attacker who has create or update permissions on configuration objects to craft labels that manipulate Kyverno’s policy generation process. The result is the unintended cloning of Secrets from protected namespaces into the attacker’s own namespace, effectively exposing sensitive data and violating confidentiality. The weakness is a classic example of CWE-639, where authorization checks are improperly handled.

The affected product is Kyverno, a Kubernetes policy engine. Any installation of Kyverno using a version earlier than 1.19.2 is vulnerable. The vulnerability applies to all environments where Kyverno is enabled and where users possess create/update rights on ConfigMaps or similar resources.

The CVSS score of 8.4 classifies this flaw as high severity, indicating significant impact if exploited. The EPSS score is not available, making it unclear how frequently attackers might target this specific path, but the lack of KEV listing suggests no widespread, confirmed exploitation yet. The attack requires an authenticated user with the ability to create or update configuration objects; once that premise is met, an attacker can trigger the Generation Handler by labeling a ConfigMap to emulate internal Kyverno labels. The resulting secret duplication represents a direct confidentiality breach and potentially grants further lateral movement opportunities if secrets contain credentials or keys.