BigQuery column security via policy and governance tags

BigQuery offers fine-grained access to sensitive columns through policy tags based on Data Catalog or data governance tags based on Resource Manager tags. Classification drives policies that check at query time whether a user may see a column—for example requiring membership in group:high-access for TYPE_SSN columns. Dynamic data masking can further substitute null, default, or hashed content for the real value.

The documented workflow is: define a taxonomy and policy tags; assign tags to columns via schema annotations; enforce access control on the taxonomy; manage IAM on each policy tag so the policy applies to every tagged column. Column-level control stacks on dataset ACLs, so both dataset and policy-tag permission are required. Sensitive Data Protection data profiles are suggested to find what needs tagging across organization, folder, or project.

For governed meaning, the taxonomy and policy tags are the shared classification of column sensitivity, not only a lock. Schema annotations bind that classification to physical columns so access and masking follow the same labels. Data and AI teams get a single place to decide which semantic classes of fields are visible, while query engines enforce those decisions instead of relying on tool-specific filters.