Analysis

Continuous Authentication: Why Identity Doesn’t End at Onboarding

Mayte Hernández Content and communication Specialist
Published: 9 May, 2025 Updated: 20 November, 2026 9 min

The industry has done a huge job securing the front door. Banks, fintechs and insurers verify who opens an account with document checks, facial biometrics and liveness detection. Yet fraud keeps evolving, because it did not stay at the door. It can appear at any point in the digital relationship: when that person logs in again, changes their details or makes a payment.

This is the gap that continuous authentication closes. Instead of confirming identity once and treating the question as answered, it keeps checking, in context and in the background, whether the person acting now is still the person who was verified on day one.

What is continuous authentication?

Continuous authentication is the practice of verifying a user’s identity throughout a digital session and across the whole customer relationship, not only at login or onboarding. It reads signals such as the device, the network, the location and the way the user behaves, and compares them with that user’s usual pattern. While everything fits, the user notices nothing. When something doesn’t, the system can request an additional check at that exact moment.

It is different from perpetual KYC (pKYC), although both share the same logic. pKYC keeps the customer profile and its risk rating up to date over time. Continuous authentication protects each interaction in real time. Together, they replace the idea of identity as a one-off event.

Why point-in-time authentication is no longer enough

When fraud teams are asked what happens after onboarding, the answer is often uncomfortable. The data exists. The login is recorded, the change of email is recorded, the transfer is recorded. But they live in three different systems, run by three different teams, and nobody reads them together.

Each of those systems does its job well. The access system sees a correct login. The data system sees a legitimate email change. The payments system sees a transfer within limits. Each action looks normal on its own. Read together, on the same identity and within a short time window, they tell a very different story.

That is why we say the deficit is not one of data, it is one of continuity. The problem is rarely missing information. It is the lack of capacity to connect signals and understand what they mean in the context of a single identity. This is also why attacks with valid credentials are so effective: at login, everything looks right.

From a one-off check to a continuous story: the continuous identity approach

For years, identity answered two questions: who is this person, and are they who they claim to be. But trust is not built only in that first moment. In real life we keep asking: is this still the same person? Does what they are doing now fit with what they usually do?

This is the idea behind Facephi’s vision: identity is no longer a one-off check, it is a continuous story built with context. Identity becomes the element that connects different signals and makes risk easier to understand. The clearest way to see it is to follow one customer over time.

Day 1: onboarding opens the identity

A customer opens an account with a bank. A guided flow helps her capture her document and selfie correctly, detecting in real time when an image may not be valid and telling her how to fix it. For her, the experience is smooth. For the bank, it is the first step in establishing a verified identity and generating the evidence that will accompany her story.

This is also where threats have changed the most. It is no longer only about a fake document at a counter, but about deepfakes and injection attacks, where synthetic video never passes through the camera and is injected directly into the flow. Protecting this first moment requires liveness detection and specific injection attack defense. But onboarding does not close the identity. It opens it.

Month 3: accumulated context

Three months later, the same customer opens the app and makes a transfer. From the first second, the session generates signals that add context to her identity. Each layer answers a different question about the same person:

  • Is she who she says she is, here and now? Identity verification does not have to stay in onboarding. If the context requires it, it can be checked again.
  • Can we trust this interaction? Behavioral intelligence reads the session continuously: device, network and context, compared with her usual pattern.
  • What is this account and does it represent risk? Account risk looks beyond a single operation, at the account’s age, its beneficiaries and its possible role in a network of accounts.
  • And this payment? Transaction monitoring and AML read the amount, the corridor and the parties as part of that same network, not as an isolated movement.

Two indicators help read this story. One is how much risk the identity represents right now. The other is how mature the profile is, meaning how much is known about it. An account with three years of history and an account opened yesterday can show the same low risk, but the confidence behind each assessment is very different. Little history is not suspicion. It is legitimate lack of knowledge, and it deserves to be treated as such.

The Thursday that doesn’t fit

On an ordinary Thursday, someone logs in with the customer’s credentials from a different device, changes the contact email and adds a new beneficiary. In a fragmented stack, these are three normal events in three systems.

Continuous authentication changes the question. It does not look at three events, it reads a sequence on the same identity. The device, the network and the context show anomalies during the session, and the account activity adds new relationships to the picture. At that point, the only question that matters is: is this still you?

The response does not have to be a block. It can be proportional to the risk: a dynamic challenge, such as a facial verification with passive liveness, at that moment of the interaction. If the check is passed, the institution can let the operation continue. If it is not, it can apply additional measures or stop it, according to its own policies and risk appetite. Blocking a legitimate customer costs money and costs the relationship. The best anti-fraud experience is the one a legitimate customer never notices.

Weeks later: evidence that can be explained

Deciding well is not enough; the decision has to be defensible. Weeks after that Thursday, the fraud team needs to review what happened: which decision was taken, why, with which information and with which evidence. A complete identity profile brings together the activity, the interactions and the operations of that identity, and each verification is backed by a signed audit record. This traceability is what turns a model into something an auditor or a regulator can follow.

The signals behind continuous authentication

Continuous authentication works because most of its signals are collected passively, without asking the user to do anything:

  • Device and network signals: device and browser characteristics, IP risk, location and changes compared with previous sessions.
  • Behavioral biometrics: typing rhythm, navigation patterns and other micro-behaviors that are as personal as a fingerprint. Behavioral biometrics detects subtle changes that may indicate impersonation, without adding friction.
  • Facial biometrics with passive liveness: when a step-up is needed, the user only has to look at the camera. Passive liveness confirms that a real person is present, without asking for specific movements.
  • Sequence and context: the order and timing of actions on the same identity, which is where isolated signals become meaningful. This is the core of contextual defense.

Continuous authentication and regulation in the US and Canada

There is no single rule that mandates continuous authentication, but regulators in North America are moving in the same direction: identity and risk must be managed over the whole relationship, not only at the start.

  • United States, NIST: NIST SP 800-63B-4, published in July 2025, sets reauthentication limits by assurance level (for AAL2, an overall timeout of no more than 24 hours and an inactivity timeout of no more than one hour) and includes guidance on session monitoring to detect fraud during an active session.
  • United States, FinCEN: the Customer Due Diligence Rule requires financial institutions to conduct ongoing monitoring and to keep customer information up to date on a risk basis.
  • Canada, FINTRAC: under the PCMLTFA, reporting entities must carry out ongoing monitoring of their business relationships, with a frequency based on each client’s risk, and keep records of the measures taken.

Continuous authentication does not replace these obligations, but it gives them what point-in-time controls lack: up-to-date signals, a consistent view of each identity and the evidence to justify every decision.

Toward continuous identity with Facephi

Facephi Intelligence Platform is built on this idea: placing identity at the center and connecting verification, session, account and transaction to understand risk across the whole identity lifecycle. Identity verification and behavioral intelligence are available today, and account risk and AML transaction monitoring are part of the platform’s direction. It works with each institution’s current stack, and the final decision always remains with the institution, according to its own rules.

Every identity has a story. The question is not whether your platform has it, but whether you can read all of it today or only the first chapter. Onboarding is day one. Fraud can strike on any other day.

Facephi presents this vision in Tech Frontiers of the New Era: Continuous Identity and Fraud Reimagined, an on-demand event premiering on November 18. If you want to see how continuous authentication would fit into your organization, contact our team.platform. 

Continuous authentication checks a user’s identity throughout a digital session, not only at login. It reads signals such as device, network, location and behavior, and compares them with the user’s usual pattern. When something doesn’t fit, it can trigger an additional check, like a facial verification, at that point in the interaction.

Traditional authentication confirms identity once, at login or onboarding, and treats the question as closed. Continuous authentication keeps asking “is this still the same person?” during the whole relationship. Fraud often happens after onboarding, when someone accesses the account again, changes contact details or makes a payment.

Continuous authentication protects each session in real time. Perpetual KYC (pKYC) keeps the customer profile and its risk rating up to date over the whole relationship, without waiting for periodic reviews. They complement each other: one confirms who is acting now, the other confirms that what you know about the customer is still valid.

Common signals include device and browser characteristics, network and IP risk, location, and behavioral biometrics such as typing rhythm or navigation patterns. Read in isolation, each signal can look normal. Read together, on the same identity, they show whether a sequence of actions fits that customer’s history.

Not when it’s risk-based. Most signals are collected passively, so a legitimate user notices nothing while their behavior matches their usual pattern. Extra steps, such as a facial check with passive liveness, only appear when the risk justifies them. The response is proportional: allow, challenge, or block, according to each institution’s policies.

There is no single mandate, but the direction is clear. NIST SP 800-63B-4 (July 2025) sets reauthentication limits and includes guidance on session monitoring. In Canada, FINTRAC requires ongoing monitoring of business relationships, and in the US, FinCEN’s CDD Rule requires ongoing monitoring and keeping customer information current.

In account takeover, the attacker uses valid credentials, so login looks legitimate. Continuous authentication detects the sequence that follows: a new device, a changed email, a new beneficiary. Each event looks normal in its own system, but read on the same identity they reveal the risk before the money moves.

It’s the idea that identity is no longer a one-off check but a continuous story built with context. Onboarding is day one, and every later interaction adds signals and evidence to that identity, so each decision can be explained and audited weeks later.

Ready to protect your users?

Discover how Facephi's biometric technology can safeguard your identity verification process.

Facephi Facephi Identity Platform Onboarding Authentication UX Consultancy Facephi Builder Facephi Central Services Fraud Intelligence Platform Identity Fabric KYB Platform Teseo Identity Wallet IDV Suite Cuentas Mula Behavioural Biometrics Linkedin YouTube X Facebook
Secret Link