What Is Digital Identity? Identity Verification Explained

Digital identity for KYC teams: what identity verification proves at onboarding, how authentication differs at each login, and what assurance levels mean.
A digital identity is any set of attributes that uniquely describes someone within a given context, and one person can hold many. For a bank or payment firm, the identity that counts is narrower: the customer’s official identity, tied to a real person. Identity verification makes that tie at onboarding; authentication then confirms, at each return, that the person logging in is the one who was verified.
This guide draws on the Financial Action Task Force (FATF) Guidance on Digital Identity of March 2020, a non-binding text, and on US federal guidelines for terms and assurance levels. Singapore’s Singpass is the one national example. It is written for compliance, onboarding and fraud teams and product managers.
What is digital identity?
A digital identity is “an attribute or set of attributes that uniquely describes a subject within a given context.” That is the wording of NIST Special Publication (SP) 800-63-4, the US National Institute of Standards and Technology (NIST) guidelines published in July 2025. The guidelines cover users of government information systems, such as employees, contractors and private individuals.
A digital identity is unique within an online service. NIST adds that “the real-life identity of the individual behind the digital identity might not be known,” a gap that identity proofing is meant to close.
For customer due diligence (CDD), the FATF uses a narrower sense: for the FATF’s purposes, “identity” means official identity, which is separate from personal and social identity. Official identity is “the specification of a unique natural person” based on attributes or identifiers. It “is recognised by the state for regulatory and other official purposes.”
In IT security, “digital identity” also covers user accounts and machine identities handled by identity and access management (IAM). This article uses the official sense.
Attributes, identifiers and credentials
Attributes are qualities ascribed to a person; NIST’s examples are name, date of birth and address. Identifiers, such as an ID number, help establish that a person is unique in a population.
In the FATF guidance, proof of official identity “generally depends” on government-provided registration, documents or certification, such as an identity card or a digital ID credential. The criteria “can vary by jurisdiction.”
A credential links that identity to something the person holds. In NIST’s definition, it “authoritatively binds an identity — via an identifier — and (optionally) additional attributes to at least one authenticator.” The person possesses and controls that authenticator. The FATF’s examples include passwords, a one-time code generator on a smartphone, public key infrastructure (PKI) smart cards and Fast IDentity Online (FIDO) certificates.
Digital identity vs a digital ID system
An email account or a work login is a digital identity in the broad sense. A digital ID system has a narrower job. In the FATF’s words, digital ID systems “use electronic means to assert and prove a person’s official identity online (digital) and/or in-person environments at various assurance levels.”
General-purpose ID systems, such as national ID, prove official identity for all or most purposes, though the guidance notes that not every jurisdiction has one.
How a digital ID system works
The FATF guidance, following NIST’s model, describes two essential components and an optional third.
| Component | Question it answers | What happens |
|---|---|---|
| Identity proofing and enrollment | “Who are you?” | Evidence checked and tied to one person; account opened; authenticators bound |
| Authentication | “Are you the person who has been identified and verified?” | Holder proves control of authenticators bound to the account |
| Portability and interoperability (optional) | Can other organizations rely on it? | Credentials reused without repeating identification and verification |
The questions in the first two rows are the FATF’s; the third is our summary. By the FATF’s definition, proofing and enrollment can be physical, digital or both, while binding, credentialing and authentication are digital. The guidance says the first component “overlaps with” the identification and verification measure of CDD.
Inside identity proofing
The FATF lists the actions inside the first component “for the purposes of illustration only,” and their order can differ.
- Collection. The applicant supplies data and evidence, for example an online form, a selfie and photos of a passport.
- Validation. The document is inspected for authenticity, for example its security features and expiry date, and its data for accuracy.
- De-duplication. The system establishes that the attributes and evidence relate to a unique person within it.
- Verification. The person is linked to the evidence, in the FATF’s example with “biometric solutions like facial recognition and liveness detection.”
- Enrollment and binding. An identity account is created, and one or more authenticators are issued and linked to it.
In practice, one such solution compares a selfie with the document photo, a one-to-one match called face verification. Liveness detection is meant to show that the selfie comes from a live person present at capture.
Each of the five steps is also a point where a genuine applicant can drop out. Measure drop-off at each one before you add or tighten a check, so the decision rests on what the change costs in completion.
Identity verification vs authentication
NIST defines identity proofing as “the processes used to collect, validate, and verify information about a subject to establish assurance in the subject’s claimed identity.” Verification is one part of it.
In NIST’s words, verification confirms that the applicant “holds the claimed real-life identity represented by the validated identity attributes and associated evidence.” NIST calls the term “synonymous with identity verification,” while industry usage, like this article’s title, often applies “identity verification” to the whole proofing process.
Authentication is a different act. NIST defines it as “the process by which a claimant proves possession and control of one or more authenticators bound to a subscriber account.” The claimant does this “to demonstrate that they are the subscriber associated with that account.”
The FATF guidance draws the contrast from the side of the relying party, such as a bank. Proofing gives it a level of confidence, depending on its strength, that identity information, such as name and age, is correct and relates to a real person. Authentication assures it that the person presenting the credential “is really the person to whom it belongs, and not a thief or imposter.”
The FATF names three types of authentication factors: something you possess, something you know and something you are. “A single authentication factor is generally not considered sufficiently trustworthy,” the guidance adds. A process that combines several types is “usually considered more robust and reliable.”
| Aspect | Identity proofing (verification) | Authentication |
|---|---|---|
| When it runs | At onboarding, or when re-proofing is needed | Each time the holder returns |
| What the person presents | Identity evidence and attributes | Authenticators bound to the account |
| Result | Confidence that the claimed identity is the applicant’s real identity | Confidence that the returning person is the subscriber |
The split matters for design. A strong login cannot repair weak proofing: if a synthetic identity passes onboarding, every later authentication confirms the same fraudster. Strong proofing, in turn, does not show who holds the phone six months later.
Give each flow its own owner and its own metrics. For proofing, track completion and manual review load; for authentication, track login success and how often a step-up is triggered.
Assurance levels: how much confidence is enough
An assurance level states how much confidence a process provides. The FATF glossary calls it “the level of trustworthiness, or confidence in the reliability of each of the three stages of the digital ID process.”
NIST SP 800-63-4 supersedes SP 800-63-3. It is written for US government online services, and private-sector organizations “may consider the use of these standards where appropriate.” It is not law in Southeast Asia.
Within its scope, NIST sets three separate scales of three levels each: the identity assurance level (IAL) for proofing, the authentication assurance level (AAL) and the federation assurance level (FAL). IAL is “a category that conveys the degree of confidence that the subject’s claimed identity is their real identity.”
AAL is “a category that describes the strength of the authentication process.” NIST describes AAL1, AAL2 and AAL3 as giving basic, high and very high confidence that the claimant controls the bound authenticators.
| Level | IAL (identity proofing) | AAL (authentication) |
|---|---|---|
| 1 | Core attributes validated against authoritative or credible sources; linked to the applicant | Single-factor or multi-factor authentication |
| 2 | Additional evidence; more rigorous validation and verification | Two distinct factors; approved cryptography |
| 3 | On-site attended session with a trained proofing agent; at least one biometric | Phishing-resistant public-key authenticator; non-exportable private key |
Source: NIST SP 800-63-4 §3.3.2.1 (IAL) and §3.3.2.2 (AAL), high-level descriptions; the normative requirements are in SP 800-63A-4 and SP 800-63B-4.
SP 800-63A-4 also defines a “no identity proofing” case below IAL1. The FATF guidance predates this edition, so its account of NIST levels follows SP 800-63-3.
Other frameworks keep their own scales. The FATF guidance, written in 2020, describes the EU’s electronic identification and trust services (eIDAS) Regulation as defining three levels: low, substantial and high. ISO/IEC 29115:2013 specifies four levels of entity authentication assurance. Each scale has its own owner and audience, so read a level inside the framework that defines it. The FATF guidance itself “does not require or recommend any particular assurance levels.”
Risks of digital ID and where each control sits
Many risks of digital ID also exist for documentary ID. The FATF guidance adds that proofing or authenticating people over the internet creates risks specific to digital ID systems, “particularly in relation to cyberattacks and potential large-scale identity theft.” In its glossary, a synthetic identity combines real, usually stolen, information with fake information, and impersonation means pretending to be another genuine person.
The pairing of risks, components and controls below is ours.
| Risk | Component it hits | Control in the sources |
|---|---|---|
| Synthetic identity | Identity proofing | Validation against authoritative or reliable sources; de-duplication |
| Impersonation of a real person | Identity proofing (verification) | “Biometric solutions like facial recognition and liveness detection” |
| Thief or imposter using a credential | Authentication | Several factor types; two distinct factors at AAL2 |
| Phishing of authenticators | Authentication | Phishing-resistant authenticator at AAL3 |
| Cyberattacks; large-scale identity theft | Proofing and authentication online | Assurance review of technology, governance, processes and procedures |
Sources: FATF guidance ¶10, ¶64, ¶67, ¶84, Box 1 and glossary; NIST SP 800-63-4 §3.3.2.
How digital identity fits customer due diligence
Proofing, authentication and the assurance each reaches are what the FATF guidance suggests an institution assess before it uses digital ID for CDD.
Under FATF Recommendation 10(a), financial institutions should be required to identify the customer and verify that customer’s identity “using reliable, independent source documents, data or information.” The FATF Recommendations, updated June 2026, are addressed to countries, so the binding duty comes from national law. The wider Know Your Customer (KYC) process is covered in our guide to KYC verification.
The FATF guidance reads Recommendation 10 as technology-neutral: R.10(a) “does not impose any restrictions on the form” of identity evidence, documentary or digital. The guidance also notes that the FATF standards leave it to each jurisdiction to set how a verified identity is linked to a real person. National rules are therefore the first thing to check.
In the guidance’s reading, “reliable, independent” means that the digital ID system relies upon “technology, adequate governance, processes and procedures that provide appropriate level of confidence that the system produces accurate results.” Assurance levels, it adds, are useful for judging this.
What the FATF guidance suggests checking
The guidance’s suggestions for regulated entities are condensed below; the second and third items form its two-step, risk-based approach. They come from a non-binding text.
- Understand the system’s components, especially proofing and authentication, and which CDD element each supports.
- Establish its assurance levels, looking at its technology, architecture and governance.
- Decide whether those levels fit the risks of the customer, product and jurisdiction.
- Where permitted and risk is low, consider whether lower assurance is enough for simplified due diligence.
- Make sure authorities can obtain the underlying identity information and evidence.
Remote onboarding and risk
Remote onboarding is not automatically higher risk under the FATF texts. In the guidance, non-face-to-face identification through reliable, independent digital ID systems with appropriate risk mitigation in place “may present a standard level of risk, and may even be lower-risk where higher assurance levels are implemented and/or appropriate ML/TF [money laundering and terrorist financing] risk control measures, such as product functionality limits …, are present.”
The Interpretive Note to Recommendation 10 (INR.10), paragraph 15(c), lists non-face-to-face relationships or transactions as a potentially higher-risk example. It does so only “where appropriate risk mitigation measures have not been implemented,” and its examples are not mandatory. Remote onboarding in Southeast Asia is covered in our guide to electronic KYC (eKYC).
Where authentication fits CDD
At onboarding, the guidance treats authentication as relevant in one case: the customer opens the account with digital ID credentials already issued by a provider. The provider did the proofing earlier; the institution authenticates the credential and relies on that proofing.
After onboarding, the guidance encourages institutions that authenticate existing customers through digital ID systems to use that data for ongoing due diligence and transaction monitoring under R.10(d).
Map your own flow
This map is illustrative; fill the bracketed fields from your national rules and risk assessment.
| Step | Component | What to record |
|---|---|---|
| Document capture and checks | Proofing: collection, validation | [assurance target]; [your national eKYC rule]; completion at step |
| Selfie match and liveness | Proofing: verification | [assurance target]; [rule]; completion and retries at step |
| Account opening | Proofing: enrollment and binding | [authenticators bound]; [rule]; completion at step |
| Login and step-up | Authentication | [assurance target]; [rule]; login success; step-up rate |
A national digital ID in practice: Singpass
Singpass is in operation, and the Monetary Authority of Singapore (MAS) calls it “Singapore’s national digital identity” on its Digital ID and e-KYC page (last updated December 2021). The Government Technology Agency (GovTech) describes it as “a trusted digital identity that allows residents to access government and business services online,” on a page updated on 15 September 2026. Login uses “biometrics (fingerprint, Face Verification) or SMS Two-Factor Authentication (2FA).” After login, participating services can pre-fill forms with Myinfo data, and MAS lists customer onboarding with Myinfo, with explicit consent, among uses by financial institutions.
The following mapping to the FATF method is ours. Read that way, two cases appear. Myinfo data shared at onboarding can be read as identity evidence for identification and verification under R.10(a). A Singpass login used to open an account can be read as the FATF’s case of authentication with credentials the customer already holds. Before relying on either, an institution still checks what MAS allows and whether the assurance fits the risks involved.
Sources checked on 10 November 2026. This article is for informational purposes only and does not constitute legal or regulatory advice. The FATF guidance is non-binding, and NIST SP 800-63-4 is a set of US federal guidelines. Organizations should confirm how identification and verification requirements apply to them with the relevant authority or legal counsel.
Behind every digital identity a regulated business relies on sit two questions with different answers and different owners. Proofing asks, usually at onboarding, whether the attributes belong to a real person and whether the applicant is that person. Authentication asks, at every return, whether the person in control of the bound authenticators is that same person.
Assurance levels put a measure on each answer, on scales that are separate by function inside NIST and by framework outside it. The FATF’s non-binding guidance gives a method for judging a system without naming a target level, which leaves the target to national rules and your own risk assessment.
For a team that owns onboarding and login, the next step is to fill in the flow map with your own assurance targets, national rules and metrics. Once filled in, it shows where a rule sets the level and where the choice is yours.