Fiche 16/54 Part 2 — Core Principles Intermediate Reviewed 2026-08-23

Anonymisation vs Pseudonymisation (Art. 4.5)

Pseudonymisation is a technical security safeguard (the GDPR fully applies). True anonymisation permanently prevents re-identification (the GDPR ceases to apply).

Open in interactive reader

⚖️ Legal Boundaries 📋 Recital 26 & WP29 Criteria

Anonymisation vs Pseudonymisation (Art. 4.5)

⚡ In 30 seconds

Under Article 4(5), pseudonymisation replaces direct identifiers with artificial pseudonyms, where re-identification remains possible via separately stored secret keys. Pseudonymised data is always personal data. By contrast, anonymised data can never be linked back to an individual by any reasonable means likely to be used; it entirely exits the scope of the GDPR.

Criterion	Pseudonymisation (Art. 4.5)	Anonymisation (Recital 26)
GDPR Scope	Applicable in full: DPA, user rights, data breach notification.	Excluded: Regulation ceases to apply.
Reversible?	Yes, by using separate key or lookup attribution tables.	Irreversible across all state-of-the-art technological means.
Role in Compliance	Security measure (Art. 32) & Privacy by Design (Art. 25).	Definitive exit strategy for open-data publishing.

⚡ Key Takeaways

Three test thresholds: singling out, linkability, and statistical inference.
Attribution information must be kept separately and subject to technical measures.
The process of converting personal data into anonymous records is itself a processing operation.

⚠️ Common Pitfall

Publishing 'anonymised' research tables that can be trivially re-identified by cross-referencing postal codes and birth years.

🛠️ In Practice

Store pseudonymisation secret keys in dedicated Hardware Security Modules (HSMs) or isolated Key Vaults.

⚖️ Official Sources: Art. 4(5) GDPR · Recital 26 · WP29 Opinion 05/2014 on Anonymisation Techniques ✓ Legal review: 23 August 2026