Is your data really anonymous?
Deleting the name column is not anonymization.
It feels safe: strip the names, maybe the emails, and share the file. But under the GDPR the test is not "did I remove the obvious identifiers?" — it is "can this person still be singled out?" Very often the answer is yes, because the details left behind still point to one individual. This page walks through why, with the primary sources.
The legal test: reasonably likely to identify
GDPR Recital 26 defines the boundary. Data is anonymous only when a person is not or no longer identifiable, taking into account all the means reasonably likely to be used to single them out — by anyone, not just you. If re-identification remains reasonably possible, the data is still personal data and the GDPR still applies, no matter how many name columns you deleted.
Why leftovers identify people: quasi-identifiers
A quasi-identifier is a field that is not unique on its own but becomes identifying in combination. None of these are "names" — yet together they single people out:
| Combination left in the data | What it does | Source |
|---|---|---|
| ZIP code + date of birth + sex | Uniquely identifies ~87% of the U.S. population | Sweeney, 2000 |
| Date of birth + sex alone | Encoded inside every PESEL number | PESEL structure |
| Rare job title + small town | Often points to a single person | Small-cell risk |
The evidence, in primary sources
GDPR Recital 26
Data is only anonymous if a person is "not or no longer identifiable" — accounting for all means reasonably likely to be used to single them out. If re-identification stays reasonably possible, the data is still personal data.
The 87% study
Latanya Sweeney showed that the combination of 5-digit ZIP code, date of birth and sex uniquely identifies roughly 87% of the U.S. population — none of which is a "name."
Sweeney, L. (2000). Simple Demographics Often Identify People Uniquely. Carnegie Mellon University.
k-anonymity
A dataset is k-anonymous if every record is indistinguishable from at least k−1 others on its quasi-identifiers. When k = 1, a record stands alone and the person is re-identifiable — deleting the name column does not raise k.
Sweeney, L. (2002). k-anonymity: a model for protecting privacy.
k-anonymity, in one sentence
A record is protected only when it hides in a crowd: at least k people share the same quasi-identifiers. When only one person matches (k = 1), that record is re-identifiable — and deleting the name column does nothing to raise k. Real anonymization means removing or generalizing the quasi-identifiers until the crowd is big enough, not just erasing the obvious fields.
What to do about it
Treat quasi-identifiers as first-class targets: date of birth, postcode, rare job titles, employer, and structural identifiers like a PESEL (which encodes date of birth and sex in its digits). Amaze detects and masks personal data — including Polish identifiers and names across grammatical cases — locally, on your own machine, so the fields that actually re-identify people don't slip through.