Biometric data definition under privacy laws
Key takeaways
- Whether data counts as biometric depends on the applicable law, since Illinois, Texas, California, and the GDPR each define it differently.
- A facial template stored as numbers can still be biometric data, while photographs are excluded under Illinois BIPA and need separate analysis.
- Under the GDPR, biometric data processed to uniquely identify a person needs both an Article 6 lawful basis and an Article 9 condition.
- BIPA's destruction rule uses the earlier of purpose satisfaction or three years after last interaction, so a vague keep-until-no-longer-useful retention policy can miss that deadline.
Biometric data generally means information about physical or behavioral characteristics that enables identification, but its legal definition depends on the applicable law. Fingerprints, facial geometry, and voiceprints are common examples; ordinary photographs and recordings require closer analysis. This guide explains the distinctions across selected U.S. laws and the EU GDPR, then translates them into practical classification and protection steps. (ilga.gov)
What counts as biometric data under privacy laws?
Biometric data can include measurements, scans, or information derived from a person’s biological or behavioral characteristics. Some laws list specific identifiers, while others focus on technical processing and whether information establishes someone’s identity. The biometric data definition under privacy laws therefore turns on the jurisdiction, the information involved, and sometimes its intended use—not simply whether a product is marketed as “biometric.” (ilga.gov)
A useful distinction separates the original sample from the identifying information extracted from it. A camera may capture a facial image, after which software extracts features and stores a reference template for matching. That template can be a string of numbers rather than a recognizable picture, yet still enable identification. (ico.org.uk)
Common examples to investigate include:
- Fingerprint and palm systems: Measurements used for entry, attendance, or identity verification.
- Facial recognition: Facial features or geometry used to recognize or authenticate someone.
- Voice recognition: A voiceprint used to distinguish speakers, rather than merely transcribe speech.
- Behavioral recognition: Patterns such as gait or typing rhythm that can identify someone, depending on the law. (ilga.gov)
How major privacy frameworks define biometric information
These examples illustrate why a single global classification is unreliable. Illinois lists particular identifiers and exclusions, Texas regulates specified identifiers captured for commercial purposes, and the GDPR emphasizes technical processing and unique identification. Use each framework’s definition before assessing its obligations. (ilga.gov)
Illinois: identifiers and information derived from them
The Illinois Biometric Information Privacy Act, or BIPA, defines biometric identifiers as retina or iris scans, fingerprints, voiceprints, and scans of hand or face geometry. It separately covers information based on a biometric identifier used to identify an individual, regardless of how that information is captured, converted, stored, or shared. A numerical template is therefore not automatically outside the law. (ilga.gov)
BIPA expressly excludes photographs, written signatures, demographic information, and certain medical or healthcare-related information, among other exclusions. Those exclusions require careful reading: do not assume that a photograph and every subsequent facial-processing operation receive identical treatment. Assess the actual output against the statutory definitions. (ilga.gov)
Texas: specified identifiers and commercial capture
Texas’s Capture or Use of Biometric Identifier law covers retina or iris scans, fingerprints, voiceprints, and records of hand or face geometry. Subject to applicable exceptions, commercial capture requires informing the individual and obtaining consent beforehand. The law also addresses disclosure, reasonable protection, and destruction after the collection purpose expires. (statutes.capitol.texas.gov)
Purpose and exceptions matter here. Amendments effective January 1, 2026, include provisions addressing artificial intelligence and publicly available media. Do not assume either that public availability supplies consent or that every AI-related use falls within the same rules; review the current statutory conditions. (statutes.capitol.texas.gov)
California: identity-related characteristics and sensitive information
California’s CCPA framework distinguishes personal information from sensitive personal information. The California Attorney General identifies biometric information processed to identify a consumer as sensitive personal information. That classification matters because additional restrictions and a right to limit certain uses or disclosures can apply, subject to statutory conditions. (oag.ca.gov)
Avoid treating “sensitive” as a universal synonym for “consent required.” California’s framework provides several distinct rights and obligations rather than copying BIPA’s collection requirements. The business’s coverage, processing purpose, and applicable exceptions must be assessed separately. (oag.ca.gov)
European Union: technical processing and unique identification
Under GDPR Article 4(14), biometric data results from specific technical processing of physical, physiological, or behavioral characteristics that allows or confirms unique identification. Article 9 adds special-category protection when biometric data is processed for the purpose of uniquely identifying a person. These are related but distinct tests. (eur-lex.europa.eu)
For that special-category processing, an organization needs both an Article 6 lawful basis and an applicable Article 9 condition. Explicit consent is one possible condition, not a blanket justification for every deployment. Other conditions have their own requirements, and satisfying one does not remove the remaining GDPR duties. (eur-lex.europa.eu)
When do photographs and recordings become biometric data?
A photograph or voice recording is not automatically biometric data under every privacy framework. The key issue may be whether specific technical processing enables unique identification, or whether the system produces an identifier named in the relevant statute. Even when biometric rules do not apply, identifiable media may remain protected by broader personal data laws. (ico.org.uk)
Consider these practical distinctions:
| Activity | Classification issue to investigate |
|---|---|
| Storing employee profile photos | Whether the photos are ordinary personal information or undergo identifying technical processing. |
| Creating facial templates for building entry | Whether facial geometry or identifying biometric processing triggers specific protections. |
| Recording customer calls | Whether the system merely preserves speech or creates speaker-identification voiceprints. |
| Measuring typing patterns | Whether behavioral patterns identify individuals under the applicable definition. |
These are assessment prompts, not universal legal conclusions. Document the processing steps and compare them with each relevant law’s wording. (ico.org.uk)
Rights and obligations after classification
Classification determines which requirements need examination; it does not complete compliance. Under BIPA, covered private entities generally need advance written disclosures and a written release before obtaining covered data. They must also maintain a public retention policy and follow statutory restrictions on profit, disclosure, protection, and destruction. (ilga.gov)
BIPA’s destruction rule uses the earlier of purpose satisfaction or three years after the individual’s last interaction, subject to the statute’s provisions. This illustrates why a general “keep until no longer useful” policy can miss an explicit legal deadline. Build retention rules around the actual applicable requirement. (ilga.gov)
Data privacy rights also vary. California consumers may have rights to know, delete, correct, opt out of sale or sharing, and limit certain uses of sensitive personal information. Exceptions apply, so a deletion request does not necessarily require removing every record in every circumstance. (oag.ca.gov)
A practical classification and review workflow
Use the following checklist to organize an internal review. It is a working method, not a substitute for jurisdiction-specific legal advice.
- Describe the operation. Record what is captured, what features are extracted, and how matching works.
- Separate each data layer. Inventory source images, recordings, templates, matching outputs, and account links rather than using one broad “biometrics” label.
- Identify relevant jurisdictions. Have counsel assess coverage, affected individuals, organizational roles, and exclusions.
- Document the purpose. Distinguish authentication, identification, attendance, fraud detection, and other proposed uses.
- Map duties to controls. Assign owners for notices, permissions, rights requests, vendor restrictions, security, and deletion.
- Review changes before deployment. Reassess new matching features, additional datasets, or secondary uses instead of relying on the original approval.
This approach connects technical evidence to the legal tests. Regulators’ guidance likewise emphasizes understanding recognition processes, documenting responsibilities, and establishing clear retention periods. (ico.org.uk)
Biometric data security and ongoing oversight
Biometric data security deserves particular care because bodily characteristics generally cannot be replaced like passwords. Converting an image into a template does not remove the need for protection. Assess whether retaining the original sample is necessary, restrict access, and protect templates against misuse or reconstruction. (ico.org.uk)
A practical security review should examine encryption, access controls, vendor safeguards, deletion, and regular testing. Match controls to realistic threats and the consequences of compromise, rather than assuming that a vendor’s general security statement addresses biometric risks. Include organizational measures such as clear responsibilities and coordinated incident handling. (ico.org.uk)
Track privacy law updates alongside system changes. Texas’s 2026 amendments show how exceptions and processing purposes can alter the analysis without changing the core list of identifiers. Keep a dated record of the statutory text, guidance, and technical assumptions used in each assessment. (statutes.capitol.texas.gov)
The practical takeaway is to classify the data, its processing, and its purpose together. Biometric privacy laws and broader data protection regulations can impose different requirements on the same system. Start with the applicable definition, then evaluate permissions, data privacy rights, retention, and safeguards before collection begins. (ilga.gov)
Related guides
- How to remove metadata from a photo before sharing it
- Point-to-point vs end-to-end encryption
Sources
- ilga.gov: Fulltext
- ico.org.uk: Biometric recognition
- Texas Constitution and Statutes
- California Consumer Privacy Act (CCPA) | State of California - Department of Justice - Office of the Attorney General
- California Consumer Privacy Act (CCPA) | State of California - Department of Justice - Office of the Attorney General
- Regulation - 2016/679 - EN - gdpr - EUR-Lex
- ico.org.uk: What is special category data
- ilga.gov: 074000140K15
- ico.org.uk: Employment practices and data protection monitoring workers 1 0
- ico.org.uk: How do we keep biometric data secure