The open-source FARx library uses GetUserMedia - a native W3C standard - to capture short audio and video clips at configurable intervals throughout the authenticated session. Fully inspectable by your security team.
Clips are posted to your own server - never directly to FARx. Your server forwards them to the FARx REST API using a secure token that never touches the browser. Works with Node.js, .NET, Java, Python, PHP, Ruby, and any other server-side stack.
Every clip is processed by a multi-modal ML engine as part of the broader identity verification process, fusing voice and facial biometric signals into a continuously updated trust score. Biometric verification can include facial recognition and fingerprint validation, while FARx uses voice and facial signals in-session. Protective actions are returned in real time and applied instantly by the browser library.
Every capture is evaluated simultaneously across all five dimensions - fused into a single trust score and acted on in real time, without user intervention.
Voice and facial biometrics are continuously matched against the enrolled profile to keep verifying the customer's identity throughout the session. A mismatch triggers a protective action within the next capture cycle - typically within 10 to 30 seconds. This kind of continuous identity verification helps build a verified identity over time and strengthens customer trust in secure sessions.
Even a fraudster who seizes a valid session token is detected within seconds. Biometric re-authentication is continuous, which helps protect returning customers without repeated visible prompts. This ongoing identity authentication goes beyond a one-time login check.
Real faces are distinguished from video playback and AI-generated deepfakes through micro-expression analysis, facial depth cues, and temporal consistency checks that synthetic video cannot replicate, strengthening fraud detection. FARx also helps confirm a real person is present. For business teams, Gen-AI-related fraud affected 23% in 2024 and 35% in early 2025. AI improves fraud detection by analysing behavioural patterns, and conflicting identity signals or simultaneous physical and digital intelligence signals can raise fraud risk by 20x. Layered checks also help detect synthetic identities, not just deepfake media.
AI-generated voice clones are identified through acoustic fingerprinting - spectral irregularities and vocoder artefacts imperceptible to the human ear, with checks for resistance to injection attacks in voice streams, but reliably detected at the signal level.
FARx counts the number of faces visible in the camera frame. A second face triggers an immediate screen blur - protecting sensitive on-screen data from shoulder-surfers without alarming the authorised user or ending their session.
In an era of remote working, hot-desking, and mobile access, employees routinely access sensitive systems in environments where others can observe their screen. GDPR, HIPAA, and sector-specific frameworks require organisations to take reasonable steps to prevent unauthorised persons from viewing personal or confidential data.
FARx automates that obligation. When a second face enters the camera frame, the screen blurs immediately. When the bystander leaves, the screen clears automatically. No action required from the user. No session interrupted.
Family members entering the room trigger an instant blur of patient records, HR data, or financial information.
A passing colleague entering camera view blurs client portfolios and confidential data before they can read anything.
Anyone sitting adjacent to the user in a café, airport, or train triggers automatic screen blur - no privacy screen required.
Privileged documents are automatically obscured when an unauthorised person enters the frame - satisfying confidentiality obligations without manual action.
When FARx detects a trust anomaly, the appropriate protective action is applied instantly - calibrated to the severity of the detected threat and the risk configuration of your application.
Applied on a slight trust dip. Content blurs while FARx continues to assess - clears automatically if the score recovers, with no user action required.
Interactive controls disabled with a brief, non-alarming prompt. Suitable for persistent low confidence where the cause may be environmental rather than fraudulent.
User redirected to an explicit verification step - spoken passphrase, OTP, security questions, or supervisor override - before the session resumes.
High-confidence fraud triggers immediate logout, session token invalidation, and a fraud event dispatched to the configured webhook. No data remains exposed.
For anomalies that warrant monitoring rather than immediate action- a real-time alert is dispatched to your security or compliance endpoint while the session continues.
The HTML5 integration code works identically across all environments; only the protective-action configuration differs by context.
Continuous MFA meeting PSD2 SCA requirements without adding friction. Financial institutions use continuous website identity verification to verify customers across the session, helping secure access to a bank account and protect every transaction and account change throughout it, rather than relying only on login. These checks help prevent fraud, support AML regulations, and help businesses avoid fines and sanctions.
Ensures only the authorised employee or partner accesses the session from start to finish - blocking both shared credentials and phishing-obtained logins. Continuous verification also helps onboard customers into sensitive internal or partner systems without relying only on static login checks.
Continuously verifies that the person accessing a patient record is the authorised practitioner, detecting device sharing and idle session hijacking under HIPAA and NHS DSPT requirements.
In many UK contexts, digital identity verification is legally required for businesses and helps citizen-facing portals reach higher identity assurance tiers under UK Gov.UK One Login, eIDAS, and NIST SP 800-63 while remaining accessible and frictionless for citizens. GOV.UK One Login can use the login app to complete checks in under 2 minutes 30 seconds. Users may prove who they are with a digital ID or in person at a Post Office, depending on identity verification requirements.
Detects the biometric mismatch that occurs when a fraudster uses stolen credentials - even after successfully passing all static authentication checks including MFA OTP challenges. A smoother customer experience depends on using the right identity validation approach for online identity checks. Combining continuous checks with ID verification makes it easier to approve genuine customers. AI-powered verification reduces manual review, lowers operational costs, and speeds up customer onboarding.
Passive, continuous verification detects account sharing and impersonation without disruptive challenges that would harm legitimate player experience or platform engagement, while supporting a seamless experience for a returning user.
FARx runs as a Docker container across cloud, on-premises, private cloud, and hybrid environments - ensuring biometric data stays within your own infrastructure when required.
FARx managed. Fastest path to production. Suitable where strict on-premises requirements do not apply.
Full deployment within your own infrastructure. No biometric data, raw media, or audit logs leave your perimeter.
Dedicated tenancy on AWS, Azure, or GCP. Cloud scalability with full biometric data sovereignty. Anyone sitting adjacent to the user in a café, airport, or train triggers automatic screen blur - no privacy screen required.
Core processing on-premises with cloud burst capacity for peak traffic periods. FARx can be introduced through a single integration that fits existing workflows across mixed environments. This hybrid approach can also support regulatory compliance across mixed environments.
The decision to provide the browser integration as open-source code is fundamental to how FARx approaches the developer and security communities - enabling inspection, audit, and approval without waiting for vendor documentation.
Most teams go from first contact to a working proof of concept in under a week - no sales process required.
Explore the FARx API