The core vulnerability in CVE-2026-18621 is a confused deputy problem: the V1 API path in DSP was designed as a convenient submission shortcut for pipeline execution, but it bypasses the namespace-scoped authorization checks enforced elsewhere in the platform. Namespace editors — a role typically granted to multiple users for collaborative pipeline work — can exploit this bypass to invoke cluster-scoped pod creation, achieving node-level code execution from a relatively low-privilege starting point. The privilege escalation chain runs: namespace editor → V1 API bypass → cluster-admin pod creation → arbitrary code on the worker node. From there, lateral movement to other workloads, secret theft from mounted service account tokens, and persistence via DaemonSet deployment become trivially accessible. The CVSS 7.6 score significantly understates the actual risk by measuring only the initial node compromise without accounting for cascade probability. In a Kubernetes environment, node-root access is a gateway to cluster-wide compromise, not a contained incident. Analysts should treat this as a 9.x blast radius regardless of the official severity. The remediation priority is straightforward: the V2 API path already implements proper namespace-scoped validation, which means the secure alternative exists. The real question is why V1 was allowed to persist after V2 shipped — this is a maintenance discipline failure, not merely a code review oversight at the time of the original commit. Organizations should audit their DSP deployments for any remaining V1 API traffic, enforce deprecation timelines for legacy API paths, and specifically verify that namespace-editor role assignments are logged with the same rigor as admin-role activity. The assumption that namespace editors represent limited blast radius is what this CVE invalidates — that threat model assumption itself needs remediation.