Skip to content
atlas

Personal data

Also known as: personal information

Any information about a living person who can be named or traced, directly or by putting pieces together.

Draft - this entry has not been reviewed yet.

Formal

Under GDPR Article 4(1), any information relating to an identified or identifiable living person, directly or by combining details such as a name, ID number, location or IP address; Article 9 sets stricter rules for special categories like health, religion and union membership.

In plain English

Anything that points to you - not just your name, but your number plate, your face on a camera or the fact that "the only redhead on the third floor" called in sick.

In practice

The developer at a Danish web shop assumes the order logs are harmless, but they hold IP addresses, delivery addresses and buying habits, so the logs fall under GDPR and need access limits and a date by which they are deleted.

Why it matters

Whether information counts as personal decides whether GDPR applies at all; indirect cases like logs and IDs from devices are easy to overlook, and that is where many breaches and fines start.

Technical deep dive

GDPR Article 4(1) defines personal data through four elements: "any information", "relating to", "an identified or identifiable", "natural person". Each is read broadly. Information can be objective or subjective, true or false. It relates to a person by content, purpose or effect, so a car's telemetry relates to its driver when used to assess driving behaviour. A person is identifiable when they can be singled out directly or indirectly by reference to an identifier such as a name, identification number, location data or online identifier, or to factors specific to their physical, physiological, genetic, mental, economic, cultural or social identity. Legal persons are outside the definition (recital 14), and so are the dead (recital 27), although member states may extend protection; Denmark's Data Protection Act (databeskyttelsesloven) § 2(5) applies the rules to deceased persons for ten years after death.

Identifiability is a risk test, not a technical absolute. Recital 26 asks whether identification is possible using "all the means reasonably likely to be used", considering cost, time and available technology. In Breyer (C-582/14, 2016) the CJEU held that a dynamic IP address can be personal data for a website operator that has legal means to obtain the subscriber's identity from the ISP. In EDPS v SRB (C-413/23 P, 4 September 2025) the Court confirmed that pseudonymised data remains personal data for the controller holding the key but may not be personal data for a recipient that cannot reasonably re-identify anyone, making the assessment dependent on whose perspective is taken. Pseudonymisation under Article 4(5) is therefore a security measure, not an exit from the GDPR; only anonymisation that withstands singling out, linkability and inference, the three tests of the Article 29 Working Party's Opinion 05/2014, takes data out of scope.

Special categories under Article 9(1) are racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, genetic data, biometric data processed to uniquely identify a person, health data and data on sex life or sexual orientation; processing is prohibited unless an Article 9(2) exception applies. Criminal convictions and offences are handled separately under Article 10. Danish CPR numbers are not a special category but have their own processing rules in databeskyttelsesloven § 11, based on the opening in GDPR Article 87.

In engineering practice the hard cases are indirect identifiers: device IDs, cookie IDs, advertising IDs, precise location traces, hashed e-mail addresses (a deterministic hash of an e-mail address is a stable identifier, not anonymisation), free-text fields, log files and embeddings or model weights that can reproduce training examples. Data classification and records of processing (Article 30) should treat such fields as personal data by default, with retention limits and access control to match.

Relationships

A kind of
Asset

Sources & further reading

Official documentation

Where this data comes from

This entry was drafted by an AI from the sources above and has not yet been checked by a person. Treat it as a starting point, and check anything important against the sources.

See the review queueSuggest a correction on GitHubThis term as JSON

Mentioned in

Check yourself

Loading…

Atlas is in beta.