Personal information
CCPA‑style personal data identifying or linked to consumers or households.
How the laws define this
Every cited law and guidance document agrees that personal information fundamentally centers on data that identifies, relates to, or can be linked to an identifiable individual or natural person. However, these frameworks diverge significantly regarding who is protected and how strictly identification is established. The Protection of Personal Information Act (POPIA) extends its definition beyond natural persons to cover identifiable juristic persons, whereas California AB 2013 expands scope to include information associated with an entire household. Conversely, statutes such as the Personal Information Protection Law of the People's Republic of China (PIPL), Law 25, the Privacy Act 2020, and the Personal Information Protection Act limit protection strictly to natural persons. Frameworks also differ on what counts as identifying data: the Data Privacy Act of 2012 requires combined information to directly and certainly identify an individual, while guidance from the Office of the Privacy Commissioner explicitly encompasses technical metadata and even fabricated data like deepfakes.
Synthesised from the 11 statutory definitions below. Each one is quoted in full, with its source.
How jurisdictions differ
What is distinctive in each definition, binding law in force first. Showing 10 of the 11 instruments that define this term; all are listed under Definitions below.
| Jurisdiction | Instrument | Status | What is distinctive |
|---|---|---|---|
| China | Personal Information Protection Law | In Force | Includes information recorded electronically or otherwise, explicitly frames it as the core object of protection triggering processing rules and individual rights. |
| California, United States | California AB 2013 — Generative Artificial Intelligence: Training Data Transparency | In Force | Adopts CCPA definition verbatim and specifically requires AI developers to disclose if training datasets contain it. |
| South Korea | Personal Information Protection Act (amendment including provisions on automated decision-making/AI) | In Force | Anchors rights like portability and refusal of automated decisions, plus obligations for processing, de-identification, and cross-border transfers. |
| Quebec, Canada | An Act to modernize legislative provisions as regards the protection of personal information (Law 25) | In Force (Amended) | Explicitly encompasses names, addresses, financial records, and health data in any format concerning a natural person. |
| Alberta, Canada | Protection of Privacy Act (Bill 33) | In Force | Enumerates examples like name, address, identifying numbers, health information, educational history, and personal opinions. |
| New Zealand | Privacy Act 2020 | In Force | Covers information whether recorded or not that can identify or be reasonably used to identify an individual, focal for Information Privacy Principles. |
| Philippines | Data Privacy Act of 2012 (Republic Act No. 10173) | In Force | Requires reasonable and direct ascertainment of identity alone or certain identification when combined with other information. |
| South Africa | Protection of Personal Information Act (POPIA), No. 4 of 2013 | In Force | Extends to identifiable living natural persons and where applicable identifiable juristic persons, listing identifiers, biometrics, health, and opinions. |
| Canada | Consumer Privacy Protection Act (CPPA) (proposed, part of Bill C-27) | Under Review | Would set scope for statutory obligations and rights like access, correction, erasure in proposed federal private-sector law. |
| New Zealand | Artificial Intelligence and the Information Privacy Principles | In Force | Expands to technical metadata like IP addresses and device IDs, plus inaccurate or fabricated data like deepfakes where identification occurs across AI lifecycle. |
What it triggers
- In California, developers must disclose whether training datasets for generative AI systems contain such personal information.source
- Under the Personal Information Protection Law, it triggers processing rules and individual rights.source
- In South Korea, it anchors rights like portability and refusal of automated decisions plus processing, de-identification, and cross-border transfer obligations.source
- New Zealand's guidance sets out that it may be present at all stages of an AI system's lifecycle.source
Definitions (11)
Adopts the California Consumer Privacy Act's definition: any information that identifies, relates to, describes, is reasonably capable of being associated with, or could reasonably be linked with a particular consumer or household; developers must disclose whether training datasets contain such personal information.
Under the CPPA portion of the bill, information about an identifiable individual (broadly capturing identifiers and other data that can identify a person) that is subject to privacy protections including consent, access, correction, de‑identification standards, and breach reporting obligations. The definition determines the scope of CPPA duties and the intersection with AIDA where AI uses personal information.
Information about an identifiable natural person, whether recorded or not, that can identify or be reasonably used to identify an individual; this definition is the focal point for the Act’s protections and the application of the Information Privacy Principles.
Defined in line with the Privacy Act 2020 as information about an identifiable individual, including traditional identifiers (name, address), technical metadata (IP addresses, device IDs), and data that may be inaccurate or fabricated (for example deepfakes or fake profiles) where a person can be identified. The guidance treats personal information as potentially present at all stages of an AI system's lifecycle.
Any information from which the identity of an individual can be reasonably and directly ascertained, or which, when combined with other information, would directly and certainly identify an individual.
Personal information means information relating to an identifiable, living natural person or, where applicable, an identifiable juristic person; it includes identifiers, contact details, biometric data, health, education, opinions and other information that can identify or be linked to a person.
Information relating to a natural person that identifies or makes the person identifiable; under the amendment this core definition anchors rights (e.g., portability, refusal of automated decisions) and the scope of processing, de-identification, and cross-border transfer obligations.
Any information recorded electronically or otherwise that identifies or can identify a natural person, either alone or in combination with other information. The PIPL frames this as the core object of protection and the trigger for its processing rules and individual rights.
Information that relates to an identifiable individual, including any data that directly or indirectly identifies a person. The CPPA uses this term to set the scope of protected data subject to statutory obligations and individual rights (access, correction, erasure).
Broadly defined as any information that relates to a natural person and allows that person to be identified, encompassing names, addresses, financial records, and health data, regardless of format.
Information about an identifiable individual, encompassing data such as name, address, identifying numbers, health information, educational history, and personal opinions, which is protected under the Act.
Not to be confused with
Personal-related information identifies a person only when combined with other data, while this term covers information that identifies alone or in combination.
Personal identifying information specifically concerns data that identifies or could identify a minor, narrowing the general scope to minors.
Questions and answers
Does personal information include fabricated data like deepfakes?
New Zealand guidance includes data that may be inaccurate or fabricated, such as deepfakes or fake profiles, if a person can be identified from it. This applies across the AI system's lifecycle.
What triggers duties under China's PIPL for personal information?
It serves as the core object of protection, triggering the law's processing rules and individual rights.
Do AI training datasets need to disclose personal information in California?
Developers must disclose whether training datasets contain personal information as defined by the CCPA.
Related Terms
Personal identifying information
Data that identifies or could identify a minor....
Processing for training
Collection and retention of personal data to train ML models....
personal-related information
Information not identifying a person alone but identifying when combined with other data....
Specific Personal Data
Sensitive categories of personal data requiring heightened protection....
special categories/sensitive data
Personal data revealing highly sensitive attributes or status....