# A selfie helps show you're really here. Your selfie is usually compared with a photo of you that the service already has, such as the one on your ID or a selfie you gave it before. It can help show that a real person is at the camera and looks like that photo. What else a service does with it belongs in its notice. You’ll usually see this when a flow asks for a live photo or short video, sometimes with an ID check and sometimes to confirm it’s still you. 10 min read. ## Opening A face feels different from a document. You can replace a passport or a card. You cannot replace your face. Being asked to place it in front of a camera therefore deserves a clear explanation of the purpose and a clear account of what happens afterward. In a common verification flow, a fresh image can help connect two pieces of information: a person participating now and a portrait already tied to a claimed identity. That portrait might be the one on your ID, or a selfie you gave the same service before. The connection is useful because a photo on file cannot show who is at the camera today. That still makes a selfie one piece of evidence, not a complete answer about who you are. The comparison has a narrow job. The organization that requested it has a much broader responsibility. > A live image can help show that a person is present. If the service already holds a reference > image, such as the portrait on your ID or a selfie you gave it before, it can also help show that > the person present is you. That can help keep an account safer. ## 01. Why a fresh image is requested A photo on file shows who you were. A selfie shows who's here now. - **Photo on file:** ID portrait or earlier selfie - The gap: often years - **Present moment:** Fresh capture · today A photo a service already has, such as the portrait on a government ID or a selfie you gave it earlier, was taken at an earlier point in time. It cannot, on its own, show that the person at the camera now is the person pictured there. A fresh capture can add evidence that someone is participating during the verification. When the flow also compares that image with the photo on file, it can estimate whether the two images appear to show the same person. Some services also use face images in other ways, for example to help stop fraud and protect the people using their service. The notice shown before collection should make those purposes clear. > A fresh image can help connect a person in the moment with the identity they claim. It does not > settle every question about identity. ## 02. Three separate questions One camera step can hide three boundaries. Terms such as _liveness_, _face match_, and _verification_ often appear together. They do not mean the same thing. **Three terms, three questions** - **Liveness:** Is a real person at the camera now? A narrow question about whether a capture is consistent with a person participating now. - **Face match:** Do two images look like the same person? For example, a selfie compared with the photo on an ID, or with a selfie given to the same service before. - **The decision:** What happens next? The requesting organization decides what to do with the outcome. Move between the questions below. The point is not how a particular service reaches an outcome. It is where each claim begins and ends. **Interactive illustration: One capture, three different questions.** Move through three scenes to see where each claim begins and ends. ### 01 The present moment **Question 01** Does the capture appear to come from a person participating now? A fresh capture can support a narrow claim about presence during the flow. The exact method varies, and the claim stops there. - **The narrow outcome:** Evidence of a present moment - **Where it stops:** Presence is about the moment. It does not necessarily establish a name or specific individual as a lone check. ### 02 The portrait pair **Question 02** Do the fresh image and a reference portrait appear to show the same person? The comparison considers a fresh image against a reference, such as the portrait on an identity document or an earlier selfie. Its outcome is an estimate of resemblance, not certainty. - **The narrow outcome:** Resemblance between two images - **Where it stops:** On its own, resemblance does not establish who someone is. ### 03 What follows **Question 03** What happens after the check returns an outcome? The requesting organization decides, based on its policies, whether you continue, verify other information, and/or receive a closer review from someone on their team. - **The narrow outcome:** An input to the organization - **Where it stops:** The face check informs one step. The organization owns the decision. These portraits are illustrations. ## 03. Why it protects you If someone else tries to get in as you, a selfie can help stop them. Someone who gets hold of your details, a copy of your ID, or a password may try to open an account in your name or get into one you already have. When a service asks for a selfie, they would also have to show a face that looks like the photo it has of you, such as the portrait on your ID or a selfie you gave it before, at the camera, right then. **The same account, two different people** - **You:** Looks like the same person. You can continue. - **Someone trying to get in as you:** Does not look like the same person. The attempt can be stopped or sent for a closer look, before it reaches your account. This is one of the ways identity verification helps protect your account and your data from fraudsters, including sophisticated ones. No check catches everything, and the organization decides what happens next. ## 04. When the result is unclear Two honest images can still disagree. Same person. Different moment. Just like the real world, a face comparison produces an estimate of resemblance. It can fail to connect two images that show the same person, or find more resemblance than the situation warrants. An outcome is evidence, not certainty. Ordinary life creates differences between pictures. Sometimes the available pictures do not support a clear comparison even when nothing is wrong with the identity being presented. Performance also differs among systems and among groups of people. Public [NIST evaluations](https://www.nist.gov/publications/face-recognition-vendor-test-part-3-demographic-effects) have documented both kinds of variation. This is why an unclear result needs a path forward rather than being treated as a complete explanation of the person. **Why two honest images can differ** - **Time:** A photo on file, like an ID portrait, may be years old. - **Appearance:** Appearance changes. - **Cameras:** Images made by different cameras under different conditions do not look identical. > **What the result does not tell you:** An unclear comparison does not explain why the images > differed. It is not an accusation, and it does not decide what should happen next. ## 05. After the camera closes The data question lasts longer than the moment. A flow may process the photo itself and information created from it for a face comparison. The image and the information derived from it can have different retention rules, so it is worth asking “what is the retention period for each category of data?” Whether either kind is retained, for how long, and for what purposes, for example for fraud prevention, depend on the requesting organization, the stated purpose, and the law that applies. When Persona powers the flow, Persona's current [privacy overview](https://withpersona.com/privacy) explains its role and how it handles personal data. The organization asking you to verify may have its own notice and retention obligations as well. ## 06. What happens next A face check ends before the real decision does. **Who owns which part** 1. **The organization:** Chooses why a selfie is part of its flow, and which methods it accepts. 2. **A provider:** May perform the comparison. 3. **The organization:** Decides what happens after an outcome. 4. **If it does not work for you:** Ask the organization about another path, such as a different method or a closer review. A technology provider may perform the comparison, but it cannot explain the organization’s purpose or promise which alternative the organization will offer. For Persona flows, start with [Persona's help center](https://help.withpersona.com/for-users/) if you are unsure where to go. Use the contact route published by the organization, and do not send a selfie or identity document through an unexpected email or message. > The comparison can contribute an outcome. The organization still owns the reason, the > alternatives, and the decision that follows. ## In closing Presence is a narrow kind of evidence. A selfie check can help show that a person is present and help connect them with the identity they claim. It cannot tell you why the organization needs that connection or what the organization will decide afterward. Those answers belong in the request, the privacy notice, and the path offered when the first method does not fit. The camera may close quickly. The responsibility around it does not. _This explainer describes general industry practice for educational purposes. It is not legal, financial, or other professional advice, and it may not fit your situation._ --- _Part of Identity Explained (by Persona). https://explained.withpersona.com/understanding-identity-verification/selfie_