Pseudonymization replaces direct identifiers such as names, email addresses, or customer IDs with artificial values. The link between each pseudonym and the real identity sits in a separate table or key that only authorized people can access. The EU GDPR defines the term in Article 4(5) and treats pseudonymized data as personal data, because someone holding the additional information can re-identify it. Data anonymization, by contrast, aims to make re-identification impossible.
A common design keeps pseudonyms consistent, so the same patient becomes P-10482 in every table. Records still join across systems, and analysts can count visits or follow a treatment history without seeing names. A stable pseudonym can still be combined with fields such as birth date and postcode to single out a person, so teams also review the attributes that remain. For AI work, consistent replacement keeps the relationships a model depends on while direct identifiers stay out of the working dataset.