Privileged Containers and Pod Security Standards in GKE

GCP Study Hub
November 1, 2026

A privileged container is a container that runs with root-level access to the node it runs on, which means it can make changes to that node. In GKE, this is a separate layer from the general container security best practices, and it deals with how much power a container gets over its node. Pod Security Standards, which are built into Kubernetes, are how you put limits on that power. For the Professional Cloud Architect exam, the points to understand are why privileged containers are risky, when they are legitimately needed, and which Pod Security Standards levels allow or reject them.

What makes a container privileged

Normally, a container only sees its own file system and processes. A privileged container gets far more. It can reach the node's devices and perform actions that regular containers are blocked from doing. In doing so, it removes the isolation between the container and its node.

That isolation is what keeps a container's problems contained. Without it, a compromised container puts the entire node at risk, along with every other workload running on that node. A problem that starts in one application is no longer limited to that application.

When privileged containers are appropriate

Because of that risk, privileged containers should be avoided for regular application workloads. Most applications simply do not need node-level access, so there is no reason to take on the exposure. This is the principle of least privilege applied at the level of a container and its node.

Privileged containers are reserved for tools that must inspect or manage the node itself in order to function, such as node monitoring agents. A monitoring agent needs to read what is happening on the node, so it needs access that a regular container would not have.

Pod Security Standards

Pod Security Standards are predefined policies, built into Kubernetes, that control which security settings pods are allowed to use. When a request to create a pod is submitted, Kubernetes checks the pod's settings against the policy before the pod is created. If the pod asks for something the policy does not allow, the request is rejected and the pod never gets created.

The standards offer three levels:

  • Privileged has no restrictions. Pods can use any setting, including everything a privileged container needs.
  • Baseline blocks known risky settings. It stops the settings that are known to allow privilege escalation, while staying flexible enough for most common application workloads.
  • Restricted enforces the strictest hardening. It follows current best practices for securing pods, such as requiring containers to run as a non-root user.

How the levels apply to privileged containers

Privileged containers are allowed only at the privileged level. Baseline and restricted both reject them. By applying baseline or restricted to the places where regular applications run, you make sure a privileged container cannot be deployed there in the first place. Because the check happens when the pod is created, a pod that requests privileged access is turned away before it ever runs.

For the Professional Cloud Architect exam, the difference between the two stricter levels is worth keeping clear. Baseline blocks the settings known to allow privilege escalation while staying flexible enough for most common application workloads. Restricted goes further and enforces the strictest hardening, such as requiring containers to run as a non-root user. Both reject privileged containers, so choosing between them is a question of how much additional hardening a workload can accept, not whether privileged containers are allowed.

Our Professional Cloud Architect course covers GKE privileged containers and Pod Security Standards alongside container security best practices and the principle of least privilege, with practice questions that drill these distinctions.

Get tips and updates from GCP Study Hub

arrow