facial recognition

in English: face recognition

basicwidespread

Facial recognition is a technology that identifies or verifies who is visible in a photo or camera image based on facial features.

In plain terms

Think of a doorman who has worked in the same office building for years: they look at your face and know whether to let you in. Facial recognition does the same with a machine, only it does not get tired and can keep even millions of faces in memory.

It is used for two types of tasks. One is verification: confirming you are who you claim to be (for example, when your phone unlocks by recognizing your face). The other is identification: determining who this person is among many (for example, when police match a recording against a database).

There is a big difference between the two cases. In the first, you usually know about it and stand in front of the camera yourself: at your phone, at the bank, or at the passport gate. In the second, you often do not know that someone compared your face with a database, which is why most debate and regulation concern this.

Moreover, your face is not a password: once it leaks, you cannot change it. For this reason, biometric data derived from facial images is specially protected by law.

How it works

Face detection: the program first locates the face in the image and crops out that section.
Alignment: rotates and scales the face to a standard position so that the eyes, nose, and mouth always fall roughly in the same place.
Creating a faceprint: a neural network converts the face into a string of numbers. This is not a photo, but a compact mathematical description of the face's characteristics.
Comparison: the system measures how close this sequence of numbers is to another stored one (verification) or to all entries in a database (identification).
Decision with a threshold: if similarity exceeds a set limit, a match is signaled. A stricter threshold yields fewer false matches but rejects legitimate users more often.

An everyday example

In the morning, you look at your phone, and it unlocks itself, then the banking app grants access with your face. At the airport, the automated gate compares the camera image with the photo stored on your passport chip, while your photo album automatically groups pictures of your grandmother.

Who is affected

everyoneThis technology manages your phone, your photos, and passport gates, and you may appear in footage from public cameras.

bank customer, investorAccount opening and mobile banking login increasingly rely on facial verification, so it is worth knowing how reliable this is. → Money

lawyer, citizenFacial data used for facial recognition is biometric data; its processing is strictly restricted by the GDPR and the EU AI Act, and you have the right to know who uses it and for what. → Law

entrepreneur, executiveIf you introduce a face-recognition access system or timekeeping, you must meet strict data protection requirements, otherwise you risk fines.

A common misconception

MythFacial recognition can definitely tell who is in the image.

In realityThe system only calculates how similar two faces are, and may err in poor lighting, from a bad angle, or with low image quality. The match is therefore a lead requiring human verification, not proof.

Limits and risks

Accuracy depends heavily on image quality: errors are far more frequent with dark, blurry, side-angle, or partially obscured faces.
Several studies have shown that some systems make errors more often for certain groups by skin color, gender or age, which can lead to wrongful suspicion.
You cannot replace your face like a password, so consider carefully which application or service you grant it to.
Weaker systems can be deceived by photos, masks, or deepfake videos if they lack liveness checks.

Similar, but not the same

face detection – Face detection only determines whether there is a face in the image and where, not whose it is.
emotion recognition – Emotion recognition does not identify the person but attempts to infer mood from facial expressions, with much less certain results.
deepfake – A deepfake fakes or swaps a face in a recording, while facial recognition identifies an existing face.

Related concepts

Broader

Computer vision, biometric identification, artificial intelligence, pattern recognition

Narrower

face recognition unlock

On the edge

face detection, liveness check, deepfake, emotion recognition, fingerprint identification, mass surveillance, Neural network

A brief history

The first semi-manual facial recognition experiments were conducted by Woodrow Bledsoe in the United States in the 1960s. Deep learning brought the current level of accuracy in the 2010s, and the general public mainly became familiar with it from 2017 onwards, through smartphones unlockable by face.

Where things stand

At the end of 2026, it is commonplace in phone unlocking, bank identification, and border control. Since February 2025, the EU AI Act has banned building facial recognition databases from facial images scraped indiscriminately from the internet or from surveillance camera footage, and it limits real-time public-space identification in law enforcement to narrow exceptions.

Status: 2026 November

In the news

Appears in 2 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Apple launches Siri AI with the $299 HomeView smart home displayOct 12, 2026 · Agent Age
Lawsuit says Meta used Facebook and Instagram images to build facial recognition and generative AI without consentSep 4, 2026 · Jog

Also known as

facial recognition · facial recognition system · face identification · face recognition technology · face recognition · facial recognition · FRT · Face ID

Sources

Wikipedia: Facial recognition system

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.