Description

A flaw was found in Data Science Pipelines. A restricted user, or tenant, can exploit an improper authorization vulnerability in the setDefaultServiceAccount function. By specifying a more privileged ServiceAccount (SA) during a CreateRun request, an attacker can bypass authorization checks. This allows the tenant to run their containers with elevated privileges, potentially leading to the disclosure of sensitive information (secrets) and the ability to execute commands within other users' pods.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "7.1",
  "pubDate": "2026-08-10T21:17:20.490Z",
  "pubdate": "2026-08-10T21:17:20.490Z",
  "executiveSummary": "An improper authorization vulnerability has been identified in Data Science Pipelines, specifically within the setDefaultServiceAccount function. This security flaw enables a restricted user or tenant to bypass standard authorization controls during a CreateRun request by explicitly supplying a more privileged ServiceAccount. The primary impact of this vulnerability involves privilege escalation, allowing authenticated low-privileged tenants to execute container workloads with elevated permissions across the cluster. The associated risk implications are severe, as successful exploitation compromises tenant isolation boundaries, potentially facilitating the unauthorized disclosure of sensitive cluster secrets and enabling remote command execution capabilities within other users' pods. Attacker capabilities are constrained to users possessing restricted tenant access who can interact with the Data Science Pipelines API. Exploitation requirements mandate that the attacker has authenticated access to initiate pipeline execution requests and possesses knowledge of, or the ability to guess, the names of higher-privileged ServiceAccounts present within the environment. Affected systems include deployments of Data Science Pipelines utilizing the vulnerable function.",
  "technicalDetails": "The vulnerability stems from an improper authorization flaw residing in the setDefaultServiceAccount function of Data Science Pipelines. The root cause of the issue is the failure of the pipeline execution handling logic to adequately validate whether the submitting user or tenant possesses the necessary authorization to bind a requested ServiceAccount to their execution run. Rather than enforcing strict access control lists or validating that the supplied ServiceAccount matches the context and privilege boundaries of the requesting tenant, the application blindly accepts user-supplied ServiceAccount parameters provided during the CreateRun request. The vulnerable component is the API execution module responsible for parsing run creation payloads and assigning execution identities. To execute an attack, a restricted user crafts a malicious API request targeting the CreateRun endpoint. Within the request payload, the attacker overrides the default configuration by specifying a privileged ServiceAccount identifier in the relevant parameter handled by the setDefaultServiceAccount function. Upon submission, the backend processes the request without performing proper authorization validation against the execution context of the requesting tenant. Consequently, the pipeline controller schedules and instantiates the associated container pods utilizing the administrative or high-privilege ServiceAccount supplied by the attacker. Post-exploitation impact and payload behavior include the execution of tenant-controlled containers possessing elevated Kubernetes permissions. With these elevated privileges, the malicious tenant can interact with the Kubernetes API to read sensitive secrets, configuration maps, and environmental variables stored across the cluster, as well as execute arbitrary commands within adjacent user pods or namespaces depending on the scope of the hijacked ServiceAccount."
}