Skip to content

CONCEPT Cited by 1 source

Purpose limitation

Definition

Purpose limitation is a core data-protection principle: personal data is only processed for explicitly stated, legitimate purposes and not further processed in ways incompatible with those purposes. The 2024-08-31 Meta Privacy Aware Infrastructure post canonicalizes the engineering formulation: "Purpose limitation, a core data protection principle, is about ensuring data is only processed for explicitly stated purposes. A crucial aspect of purpose limitation is managing data as it flows across systems and services."

Engineering challenge

At small scale, purpose limitation can be enforced by point-checking controls — if statements in code, ACLs on datasets. At hyperscale, point-checking breaks down:

  • Code audits don't compose — they need continuous re-validation as the codebase evolves.
  • ACL-based separation forces physical data duplication — each purpose gets its own dataset, exploding storage and cache cost on shared-infrastructure systems.
  • Propagation is operationally unviable — when a consumer reads from source A and writes to sink B, point-checks at B must know A's purpose; this requires "complex orchestration to ensure propagation from sources to sinks."
  • Data lineage helps but isn't sufficient — lineage tracks the graph but enforcement still happens at the point, still needing per-asset audits.

Meta's 2024 position: the only primitive that scales is information flow control (IFC) — enforcing purpose limitation at runtime as data flows, not at rest at the point of access.

The three needs IFC meets (and point-checking doesn't)

From the 2024-08-31 post's needs/problem/solution table:

  1. Programmatic control — real-time checks at code execution instead of out-of-band human audits.
  2. Granular flow control — per-request / per-function-call / per-data-element decisions on shared infrastructure, avoiding physical data separation.
  3. Adaptable and extensible control — the same data asset can carry multiple, evolving privacy requirements simultaneously.

Seen in

Last updated · 319 distilled / 1,201 read