Data Classification in Unity Catalog now scans views (Beta)

Use cases

Imagine you need to apply governance and security rules across both tables and views in Unity Catalog, but data classification previously only covered base tables. Now, the Data Classification detection engine can scan regular view columns to find sensitive data.

• Automatic scanning of views using the same detectors and classes applied to tables.
• View consolidated results in the UI and the system table.
• Increased governance coverage across virtual data objects.

How to use

To enable this feature, a workspace admin must turn on view scanning from the previews configuration.

# Initial activation is done via the workspace admin Previews page.
# Once enabled, catalogs with Data Classification active will include views in their next scheduled scan.

What happens in this code:

1. The administrator accesses the Previews page in the Databricks console.
2. Enables the Data Classification View Scanning feature.
3. The system begins including views in the incremental scan cycle of the catalogs.

Limitations and requirements

• This feature is in Beta.
• Requires a workspace admin to turn on Data Classification View Scanning from the Previews page.
• Classification tags are not automatically applied to view columns by default to prevent conflicts with column-mask or row-filter policies, unless ABAC on Views (Beta) is enabled.
• Views with unsafe definitions (such as legacy Hive metastore user-defined functions, reflect(), and java_method()) are not scanned and report an UNSUPPORTED_VIEW_DEFINITION status.
• Metric views are excluded from view scanning.