criticalOther

search-v2-operator Excessive ClusterRole Privilege Escalation

First seen Aug 22, 2026 · Updated Aug 22, 2026 · CVSS 9.9

kubernetesrbacprivilege-escalationcluster-adminopenshiftoperatoragent-relevant

The search-v2-operator, commonly deployed in Kubernetes/OpenShift environments (e.g., Red Hat Advanced Cluster Management), is provisioned with a ClusterRole granting effectively cluster-admin level permissions. This over-privileged configuration allows the operator or any workload/service account leveraging it to impersonate users, forge RBAC bindings, approve CSRs, and manage ManifestWork objects, enabling full cluster takeover.

Technical Analysis

The vulnerability stems from an overly broad ClusterRole bound to the search-v2-operator's service account, which includes 'impersonate', 'create/update' on RBAC resources, 'approve' on certificatesigningrequests, and manage permissions on ManifestWork objects used for multi-cluster orchestration. An attacker who compromises the operator's pod, service account token, or exploits a supply-chain flaw in the operator's container image could pivot to full cluster-admin access by forging RBAC policies or approving malicious CSRs to mint new trusted certificates. Because ManifestWork is used in multi-cluster management (e.g., Open Cluster Management/ACM), abuse could propagate malicious workloads across managed clusters. Organizations running AI agent orchestration platforms, RAG pipelines, or LLM tool-execution frameworks on Kubernetes/OpenShift are directly exposed if these agent workloads run in the same cluster as the vulnerable operator, since a successful privilege escalation could allow attackers to exfiltrate agent API keys/secrets, tamper with agent container images, or hijack agent compute resources for lateral movement.

Affected Systems

Clusters running search-v2-operator, including deployments as part of Red Hat Advanced Cluster Management (ACM) for Kubernetes and Open Cluster Management (OCM); affected component is the operator's ClusterRole/ClusterRoleBinding configuration bundled with the operator's Helm chart or OLM subscription.

Indicators of Compromise

  • No specific file hashes, IPs, or domains published; indicator of exposure is presence of overly permissive ClusterRole bound to search-v2-operator service account (check via: kubectl get clusterrole -o yaml | grep -A20 search-v2-operator)

Remediation Steps

  1. 1

    Audit ClusterRole Bindings

    Review the search-v2-operator's ClusterRole and ClusterRoleBinding to identify and remove impersonate, RBAC-write, CSR-approve, and ManifestWork-manage permissions not strictly required.

  2. 2

    Apply Least Privilege RBAC

    Replace the broad ClusterRole with a scoped Role/ClusterRole limited to only the resources and verbs the operator needs to function, following namespace isolation where possible.

  3. 3

    Patch to Fixed Version

    Upgrade search-v2-operator to the vendor-released version that remediates CVE-2026-70496 once available from Red Hat or the OCM project.

  4. 4

    Monitor CSR and RBAC Activity

    Enable audit logging on certificatesigningrequests approvals and RBAC object modifications to detect anomalous escalation attempts.

  5. 5

    Isolate Agent Workloads

    Ensure AI agent, LLM, or RAG pipeline workloads run in separate namespaces/clusters with network policies and RBAC boundaries isolated from operator service accounts to limit blast radius if the operator is compromised.

CVE / Advisory IDs

CVE-2026-70496

Industries Most Exposed

TechnologyCloud ServicesFinancial ServicesHealthcareGovernmentTelecommunications

Sources

Respond to this threat

Pro subscribers get a full AI-generated incident-response playbook for this threat — detection, containment, eradication, and recovery steps — plus an unlimited AI Threat Advisor for questions about your environment.