CONCEPT Cited by 1 source

Column-level access control

Definition

Column-level access control enforces data governance at the granularity of individual columns in a dataset, rather than at the table or database level. Each column carries a classification tag indicating its sensitivity level, and the execution layer enforces that only workflows with appropriate authorization can read or produce columns at each classification level.

Key properties in ML platform contexts:

• Tag propagation — as data flows through pipeline modules, classification tags propagate automatically from input columns to output columns, reducing manual effort and preventing accidental declassification
• Default-deny for unclassified — columns without explicit classification are treated conservatively (highest restriction), ensuring new data doesn't leak through
• Domain-scoped enforcement — different workflow domains (experimentation, production, ad-hoc analysis) have different column-access boundaries

Seen In

• sources/2026-06-10-atlassian-architecting-scalable-ml-platforms — ML Studio's third compliance pillar; column-level data classification with automatic tag propagation; 900k+ datasets generated with built-in access control

(Source: sources/2026-06-10-atlassian-architecting-scalable-ml-platforms)