What it is
K-anonymity measures re-identification risk by finding the smallest group of rows that share the same combination of quasi-identifiers, such as zip code, birth date, and gender.
When to use it
Privacy engineers use this before releasing or sharing a dataset, to check whether any individual can be singled out by combining a few seemingly harmless columns.
Gotchas & limits
K-anonymity alone does not guarantee privacy; a small group can still be vulnerable if all its members share the same sensitive attribute, a gap known as a homogeneity attack.