Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Pod Security Standards are three Kubernetes profiles — privileged, baseline, and restricted — applied per namespace by Pod Security Admission to reject workloads that ask for too much privilege.

What Are Pod Security Standards?
Pod Security Standards are three named profiles defined by the Kubernetes project that describe how much privilege a pod may request. Privileged allows everything. Baseline blocks the well-known privilege escalations while leaving common workloads untouched. Restricted applies current hardening practice in full. They are enforced by Pod Security Admission, a built-in admission controller that reads labels on a namespace and admits, warns about, or rejects pods accordingly.
| Control | Privileged | Baseline | Restricted |
|---|---|---|---|
| Privileged containers | Allowed | Blocked | Blocked |
| Host namespaces and hostPath | Allowed | Blocked | Blocked |
| Adding Linux capabilities | Allowed | Limited set | Only NET_BIND_SERVICE, must drop ALL |
| Running as root | Allowed | Allowed | Blocked |
| allowPrivilegeEscalation | Allowed | Allowed | Must be false |
| seccomp profile | Any | Unconfined blocked | RuntimeDefault or Localhost required |
| Volume types | Any | Any | Restricted list only |
The gap between baseline and restricted is where most of the argument happens. Baseline is a realistic target for an existing estate because it rejects host mounts, host networking, and privileged containers without demanding that every image be rebuilt. Restricted additionally requires non-root execution, dropped capabilities, and a seccomp profile, which is the right destination but usually needs image work to reach.
Enforcement is configured with labels on a namespace, in three independent modes. enforce rejects non-compliant pods, audit records a violation in the audit log, and warn returns a message to whoever applied the manifest.
apiVersion: v1
kind: Namespace
metadata:
name: payments
labels:
pod-security.kubernetes.io/enforce: baseline
pod-security.kubernetes.io/enforce-version: v1.31
pod-security.kubernetes.io/audit: restricted
pod-security.kubernetes.io/warn: restricted
That combination is the standard migration pattern: enforce what the namespace already satisfies, while auditing and warning against where you want it to be. Pinning enforce-version matters — without it the namespace tracks the cluster’s current version, so a cluster upgrade can tighten a profile and start rejecting pods that deployed cleanly yesterday.
PodSecurityPolicy was the original mechanism. It was deprecated in Kubernetes 1.21 and removed in 1.25, largely because policy selection depended on the requesting user’s role bindings, which made the result of applying a manifest hard to predict. Pod Security Admission reached general availability in 1.25 as the simpler replacement: fewer knobs, namespace scope, predictable behaviour. Clusters still running third-party PodSecurityPolicy shims should treat migration as overdue rather than optional.
Pod Security Admission deliberately does only one thing. It will not enforce an image registry allow list, require resource limits, mandate a particular label, block the latest tag, or restrict which service accounts may be used. For those rules you need a general policy engine: Kyverno, Open Policy Agent Gatekeeper, or the built-in Validating Admission Policy, which uses CEL expressions and became generally available in Kubernetes 1.30 without requiring an external webhook.
It also checks only what is declared. A pod that passes the restricted profile can still mount a sensitive secret, hold a service account with cluster-wide rights, or reach every other pod in the cluster. The profile governs the security context; everything else remains the job of role-based access control, NetworkPolicy, and workload identity.
warn and audit set to restricted, enforcing nothing. Collect the violations for a fortnight.emptyDir volumes.Both the CIS Kubernetes Benchmark and the NSA and CISA Kubernetes Hardening Guidance expect an admission-time control of this kind, and for organisations in scope of the NIS Regulations it is a straightforward piece of evidence that workload privilege is managed rather than assumed. At Privia Security, we check admission policy against what an attacker can actually deploy as part of our cloud security services.
Aim for it in application namespaces, but expect exceptions. Container network interface plugins, storage drivers, node monitoring agents, and log collectors often need host access or elevated capabilities to function at all. Put those in dedicated namespaces at the privileged or baseline level, with named owners and closer monitoring, rather than relaxing the profile across a namespace that also holds business workloads.
Nothing immediately. Pod Security Admission is an admission controller, so it evaluates creation and update requests rather than running workloads. Existing pods continue until something recreates them — a deployment rollout, a node drain, or a scaling event — at which point the new pod is rejected. That delay is dangerous, so always run audit and warn modes first to find offenders before they fail at an inconvenient moment.
If your requirements go beyond pod privilege, yes. Pod Security Admission covers only the fields in the security context and a fixed list of pod specification settings. Registry allow lists, mandatory labels, required resource limits, image signature verification, and mutation of manifests all need a general policy engine. Many clusters run both: Pod Security Admission for the baseline, and a policy engine for everything else.
No. PodSecurityPolicy was deprecated in Kubernetes 1.21 and removed entirely in 1.25, so any cluster on a current version no longer has it. If your manifests still reference it, the resources are simply ignored, which can leave you believing a control is active when it is not. Migrate to Pod Security Admission, optionally alongside a policy engine for the rules it does not cover.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.