Anonymisation, De‑identification and the Myth of “Good Enough”
In many operational environments, “anonymised” is shorthand for “we removed direct identifiers.” Names, emails and customer IDs disappear, and the dataset is declared safe for internal analytics or external sharing. But under the GDPR, this barely scratches the surface.
Recital 26 GDPR requires that data be considered anonymous only when the individual is no longer identifiable by anyone, using any reasonably available means, today or in the foreseeable future. It is a high threshold by design. If there is any plausible route back to an individual, the dataset remains personal data.
This is reinforced by the European Data Protection Board (EDPB), which has consistently held that anonymisation must withstand future technological capabilities, not merely today’s tools.
From a legal perspective, GDPR offers no sliding scale. Data is either anonymous or it remains personal data, regardless of how well protected it is.
Given this reality, organisations should adopt a more grounded default:
Treat data as de‑identified (pseudonymised), not anonymised, unless proven otherwise.
De‑identified data retains a connection to the individual, even if indirect. A key, token, or external dataset could theoretically restore identifiability. As a result, all GDPR principles continue to apply: purpose limitation, lawful basis, storage limitation, access controls, minimisation and accountability.
A practical, defensible model includes four pillars:
1) Data Quality and Accuracy: ensuring transformations and aggregations are meaningful and reliable.
2) Data Accountability and Stewardship: assigning clear ownership and decision-making authority.
3) Data Protection, Security and Compliance: ensuring technical and organisational measures remain effective.
4) Data Architecture and Lifecycle Management: designing systems that support reversible and irreversible transformations appropriately.
Without these, claims of “anonymisation” are rarely sustainable.
The recent CJEU ruling in the SRB–Deloitte case (Case C-413/23 P) makes this distinction even clearer. The Court held that pseudonymised data may be anonymous for the recipient, if the recipient has no reasonable means to re‑identify individuals.