Machine Learning workspace should use managed identity

High Azure General ml-managed-identity

Applies to

Microsoft.MachineLearningServices/workspaces

What CGPulse checks

A resource passes this rule when identity.type must be one of ["SystemAssigned","UserAssigned","SystemAssigned,UserAssigned"]. Anything else is reported as a finding with the evaluated property value attached as evidence.

How to fix it

Enable managed identity to eliminate stored credentials and improve security. In Azure Portal: Workspace > Identity > enable System-assigned managed identity. See: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-identity-based-service-authentication

CLI

az resource update --ids "{id}" --set identity.type=SystemAssigned

Terraform

resource "azurerm_machine_learning_workspace" "{name}" {
  name                    = "{name}"
  resource_group_name     = "{rg}"
  location                = var.location
  application_insights_id = "<app-insights-id>"
  key_vault_id            = var.key_vault_id
  storage_account_id      = "<storage-account-id>"
  identity {
    type = "SystemAssigned"
  }
}

Compliance frameworks

Cloud Security: Azure Foundations

Is your environment compliant with this rule?

CGPulse checks it — and 621 others — against your Azure and AWS accounts with read-only access.

Run a free scan