Role-based access control for Azure AI Foundry (hub-focused)

In this article, you learn how to manage access at the hub and project levels of Azure AI Foundry. Use Azure role-based access control (Azure RBAC) to manage access to Azure resources. Azure provides built-in roles and lets you create custom roles.

Azure AI Foundry hub vs project

In the Azure AI Foundry portal, access has two levels: the hub and the project. The hub hosts infrastructure (including virtual network setup, customer-managed keys, managed identities, and policies). It’s where you configure Azure AI services. Hub access lets you modify infrastructure, create hubs, and create projects. Projects are a subset of the hub and act as workspaces to build and deploy AI systems. In a project, develop flows, deploy models, and manage project assets. Project access lets you build and deploy AI end to end while using the hub infrastructure.

A key benefit of the hub and project relationship is that developers can create projects that inherit hub security settings. Some developers are contributors to a project and can't create new projects.

Default roles for the hub

The Azure AI Foundry hub has built-in roles that are available by default.

The key difference between Contributor and Azure AI Developer is the ability to create new hubs. Only the Owner and Contributor roles let you create a hub. Custom roles can't grant hub creation.

Azure AI Administrator role

Hubs created after 11/19/2024 have the system-assigned managed identity assigned to the Azure AI Administrator role instead of Contributor.

{
  "permissions": [
    {
      "actions": [
        "Microsoft.Authorization/*/read",
        "Microsoft.CognitiveServices/*",
        "Microsoft.ContainerRegistry/registries/*",
        "Microsoft.DocumentDb/databaseAccounts/*",
        "Microsoft.Features/features/read",
        "Microsoft.Features/providers/features/read",
        "Microsoft.Features/providers/features/register/action",
        "Microsoft.Insights/alertRules/*",
        "Microsoft.Insights/components/*",
        "Microsoft.Insights/diagnosticSettings/*",
        "Microsoft.Insights/generateLiveToken/read",
        "Microsoft.Insights/logDefinitions/read",
        "Microsoft.Insights/metricAlerts/*",
        "Microsoft.Insights/metricdefinitions/read",
        "Microsoft.Insights/metrics/read",
        "Microsoft.Insights/scheduledqueryrules/*",
        "Microsoft.Insights/topology/read",
        "Microsoft.Insights/transactions/read",
        "Microsoft.Insights/webtests/*",
        "Microsoft.KeyVault/*",
        "Microsoft.MachineLearningServices/workspaces/*",
        "Microsoft.Network/virtualNetworks/subnets/joinViaServiceEndpoint/action",
        "Microsoft.ResourceHealth/availabilityStatuses/read",
        "Microsoft.Resources/deployments/*",
        "Microsoft.Resources/deployments/operations/read",
        "Microsoft.Resources/subscriptions/operationresults/read",
        "Microsoft.Resources/subscriptions/read",
        "Microsoft.Resources/subscriptions/resourcegroups/deployments/*",
        "Microsoft.Resources/subscriptions/resourceGroups/read",
        "Microsoft.Resources/subscriptions/resourceGroups/write",
        "Microsoft.Storage/storageAccounts/*",
        "Microsoft.Support/*",
        "Microsoft.Search/searchServices/write",
        "Microsoft.Search/searchServices/read",
        "Microsoft.Search/searchServices/delete",
        "Microsoft.Search/searchServices/indexes/*",
        "Microsoft.DataFactory/factories/*"
      ],
      "notActions": [],
      "dataActions": [],
      "notDataActions": []
    }
  ]
}

Azure AI Developer role

{
  "permissions": [
    {
      "actions": [
        "Microsoft.MachineLearningServices/workspaces/*/read",
        "Microsoft.MachineLearningServices/workspaces/*/action",
        "Microsoft.MachineLearningServices/workspaces/*/delete",
        "Microsoft.MachineLearningServices/workspaces/*/write",
        "Microsoft.MachineLearningServices/locations/*/read",
        "Microsoft.Authorization/*/read",
        "Microsoft.Resources/deployments/*"
      ],
      "notActions": [
        "Microsoft.MachineLearningServices/workspaces/delete",
        "Microsoft.MachineLearningServices/workspaces/write",
        "Microsoft.MachineLearningServices/workspaces/listKeys/action",
        "Microsoft.MachineLearningServices/workspaces/hubs/write",
        "Microsoft.MachineLearningServices/workspaces/hubs/delete",
        "Microsoft.MachineLearningServices/workspaces/featurestores/write",
        "Microsoft.MachineLearningServices/workspaces/featurestores/delete"
      ],
      "dataActions": [
        "Microsoft.CognitiveServices/accounts/OpenAI/*",
        "Microsoft.CognitiveServices/accounts/SpeechServices/*",
        "Microsoft.CognitiveServices/accounts/ContentSafety/*"
      ],
      "notDataActions": []
    }
  ]
}

Default roles for projects

When you grant a user access to a project, the system also assigns the Reader role on the hub and the Inference Deployment Operator role to allow deployments in the resource group.

To create a project, a role must include Microsoft.MachineLearningServices/workspaces/hubs/join on the hub (included in Azure AI Developer).

Dependency service permissions

Sample enterprise RBAC setup for hubs

Access to external resources

Ensure the hub managed identity is granted required roles on external services (for example, storage, search) before use.

Manage access

Use the Azure AI Foundry portal (Users blade) or Azure portal IAM / CLI to assign roles.

Example CLI:

az role assignment create --role "Azure AI Developer" --assignee "user@contoso.com" --scope /subscriptions/<sub-id>/resourceGroups/<rg-name>

Custom roles

Define custom roles when built-in roles don't meet needs. Example subscription-level custom role excerpt:

{
  "properties": {
    "roleName": "Azure AI Foundry Developer",
    "description": "Custom role for Azure AI Foundry. At subscription level",
    "assignableScopes": ["/subscriptions/<your-subscription-id>"],
    "permissions": [ { "actions": ["Microsoft.MachineLearningServices/workspaces/write", "Microsoft.MachineLearningServices/workspaces/endpoints/write"], "notActions": [], "dataActions": ["Microsoft.CognitiveServices/accounts/OpenAI/*/read"], "notDataActions": [] } ]
  }
}

Assigning roles in the portal

In the management center, select Users at hub or project level, then New user.

Scenario highlights

• Customer-managed key: Grant workspace creator access to Key Vault; if using user-assigned identity, grant needed data-plane permissions.
• Connections with Microsoft Entra ID: Assign required Azure RBAC roles (for example, Storage Blob Data Contributor, Search Index Data Contributor).
• Azure Container Registry: Use system-assigned managed identity or assign ACRPull to user-assigned identity.
• Application Insights: Requires Microsoft.Insights/Components/Write and Microsoft.OperationalInsights/workspaces/write during hub creation.

Troubleshooting

If new hubs using Azure AI Administrator identity role encounter issues, you can temporarily revert to Contributor (see original article for detailed steps).

Next steps

• Create an Azure AI Foundry hub
• Create a project
• Add a connection