Two critical Red Hat Quay vulnerabilities can expose private container images or allow attackers to replace images used by downstream systems. The flaws affect registry signing keys and credentials belonging to other tenants.
CVE-2026-123710 lets an attacker install a trusted signing key and authorize access to any repository. When anonymous access is enabled, no account is required. Red Hat warns that disabling anonymous access is insufficient because any valid account can still exploit the flaw.
CVE-2026-110697 requires an authenticated user able to create a namespace. Improper wildcard handling can expose another tenant’s robot account tokens, enabling unauthorized image operations without elevated privileges or victim interaction.
For the 3.10 branch, RHSA-2026:99025, updated November 12, explicitly lists both vulnerabilities in release 3.10.28. Administrators should match their deployment to its applicable advisory. The version list in the supplied illustration comes from the robot-token CVE guidance, which also says Quay.io has been fixed for that issue.
Restrict the /keys endpoint to trusted internal components, review signing-key audit events and remove unauthorized approved keys. A fixed build is needed to prevent reinsertion. For the robot-token flaw, limiting namespace creation and rotating potentially exposed tokens reduces risk while the update is applied.
