Biometric data is information about a person's physical or behavioural characteristics, such as fingerprints, face shape, iris patterns, voice or typing rhythm, that is processed to identify a person or confirm who they are. In financial services it is used to verify new customers, to unlock phones and digital wallets, and to approve payments without a password.
The European Union's General Data Protection Regulation defines biometric data as personal data resulting from specific technical processing of such characteristics that allows or confirms the unique identification of a person, and treats it as a special category when used for identification. In payments, biometrics are the "something you are" factor in multi-factor authentication, alongside something you know, such as a PIN, and something you have, such as a phone. Biometric data describes a person; a digital fingerprint, despite the name, describes a device.
Biometric data in plain terms
A house key can be lent, copied or lost, and a password can be guessed or written on a sticky note. Your face and fingerprints are always with you and are hard to hand to someone else, which makes them convenient keys for proving who you are.
The drawback is that you cannot change them. If someone steals a password you pick a new one, but if a copy of your fingerprint or a convincing fake of your face gets into the wrong hands, you cannot get a new finger or face. That is why biometric data is guarded more strictly than most other personal information.
Types of biometric data
- Fingerprints. Used in phone sensors, some payment cards with built-in readers, and national identity systems.
- Face. Compared against an identity document photo during onboarding, or used to unlock a phone before a payment.
- Iris and palm. Iris patterns and palm veins give high accuracy and appear in some national ID programmes and bank terminals.
- Voice. Used by some call centres to recognise customers as they speak.
- Behavioural signals. Typing rhythm, swipe patterns and the way a phone is held, analysed in the background to spot account takeover.
How biometric matching works
At enrolment, a sensor captures a sample and software converts it into a template, a set of measurements rather than a photograph. Later samples are compared with the template, and the system produces a similarity score. If the score passes a threshold the person is accepted. Setting the threshold is a trade-off between false accepts, where an impostor gets in, and false rejects, where the real person is turned away.
Where the template is stored matters. In most phone-based payments, such as wallets unlocked with a fingerprint or face scan, matching happens on the device and the biometric never leaves it; the phone then signs the transaction with a cryptographic key. Passkeys built to FIDO Alliance standards follow this model and use public key cryptography so that the service never holds the biometric. Server-side matching, used in remote onboarding and national ID systems, creates large central databases that are more attractive to attackers. A liveness check tests whether a real, present person is in front of the camera rather than a photo, mask or video.
Uses in finance
During onboarding, many banks and fintechs ask for a selfie that is matched against a passport or ID card photo as part of identity verification and know your customer checks. At payment time, biometrics can satisfy the inherence element of strong customer authentication under European payment rules, often replacing a one-time password sent by text message.
National digital identity systems take this further. India's Aadhaar programme, run by the Unique Identification Authority of India, links a 12-digit number to fingerprints, iris scans and a photograph for more than a billion residents. Its biometric authentication supports electronic KYC and the Aadhaar Enabled Payment System, which lets people withdraw cash or check balances at local agents with a fingerprint, a tool often cited in discussions of financial inclusion.
Legal protection and risks
Many laws give biometric data extra protection. The GDPR generally forbids processing it to identify people unless a specific condition applies, such as explicit consent. In the United States, Illinois's Biometric Information Privacy Act of 2008 requires written consent before collecting biometric identifiers and allows individuals to sue; it led to a $650 million settlement approved in 2021 in a class action over Facebook's photo-tagging feature. Other states have their own rules, and financial firms also handle biometrics under general data privacy and data security obligations.
Fraud is the other main concern. Generative tools can now produce realistic fake faces and voices, and in November 2024 FinCEN warned financial institutions about fraud schemes using deepfake media to get past identity checks. Defences combine liveness detection, document checks and behavioural signals within broader fraud prevention programmes. Because a stolen biometric cannot be reset, regulators and standard-setters favour designs that keep templates on the user's device and use biometrics only as one factor among several.