Comparison
Data maskingvsPseudonymisation
Data masking
the analyst sees the email column as four asterisks and an at sign, while the support tool sees the real thing, from the same table.
Obscuring values at read time based on who is asking, so one table can serve different audiences. It is convenient and it is a control at the query layer, which means it protects nothing against someone with access to the underlying files. Treat it as a way to reduce casual exposure among people who already have legitimate access, not as a boundary against someone determined.
Full entry →Pseudonymisation
the user id is replaced with a hash so you can still count sessions per person, and rejoining it to a name is a deliberate act with a key.
Replacing identifiers with consistent substitutes so behaviour can still be linked per person without the person being named. It preserves the analytical joins that matter while removing the direct identifier. Most privacy regimes treat it as still personal data, because the mapping exists — which is the distinction people most often get wrong when they claim a dataset has been anonymised.
Full entry →