Identity Verification

An automated process using optical character recognition (OCR) and computer vision to extract data from identity documents and validate their authenticity. The system analyzes the document's visual security features—such as holograms, microprint, and guilloche patterns—and checks for physical tampering or pixel-level manipulation. Data extracted from the machine-readable zone (MRZ) or barcodes is cross-referenced against the visual inspection zone (VIZ) to detect alteration.

• Validates security features against known templates
• Detects digital manipulation and physical forgeries
• Extracts and cross-references MRZ, barcode, and VIZ data

The process of confirming that the person presenting an identity document is its legitimate holder. A live selfie capture is compared against the portrait on the trusted identity document using a 1:1 facial matching algorithm. This step binds the physical individual to the credential, preventing the use of lost, stolen, or borrowed documents. Advanced systems generate a similarity score based on facial geometry and landmark distance vectors.

• 1:1 facial matching against document portrait
• Generates a similarity confidence score
• Prevents impersonation with stolen documents

A critical safeguard that distinguishes a live human presenter from a spoofing artifact. Passive liveness detection analyzes the texture, micro-movements, and light reflection properties of a single image or video frame to detect presentation attacks without requiring user interaction. Active liveness prompts the user to perform a random action, such as blinking or turning their head, to defeat static masks, printed photos, and digital replay attacks.

• Passive: Analyzes image texture and sensor noise patterns
• Active: Challenges user with randomized motion prompts
• Defeats deepfakes, 3D masks, and screen replays

The systematic validation of identity attributes against authoritative third-party data sources. This includes checking the identity document number against government revocation lists, validating the person's name and date of birth against credit bureau files, and screening against global sanctions lists, Politically Exposed Persons (PEP) databases, and adverse media. This step confirms the identity's existence and standing in the real world.

• Government document validation and revocation checks
• Credit bureau and utility data cross-referencing
• Sanctions, PEP, and adverse media screening

A passive identification technique that collects unique attributes of a remote computing device to generate a persistent, probabilistic identifier. By analyzing parameters such as the browser user-agent, installed fonts, screen resolution, timezone, WebGL renderer, and TCP/IP stack characteristics, the system can recognize returning devices. This is used to detect velocity anomalies and link verification attempts to known fraudulent devices.

• Collects 100+ device and browser attributes
• Detects emulators, virtual machines, and rooted devices
• Links anonymous sessions to known fraud signals

The process of aggregating and evaluating an identity's publicly available online presence to assess its authenticity and longevity. A genuine identity typically has a digital history—social media profiles, forum posts, domain registrations—that spans years. A synthetic identity often lacks this depth or has profiles created in a coordinated batch. This analysis provides a passive risk signal that is difficult for fraudsters to fabricate convincingly.

• Evaluates social media account age and activity patterns
• Checks domain registration history and email age
• Identifies batch-created or shallow digital profiles

An automated process using optical character recognition (OCR) and computer vision to extract data from identity documents and validate their authenticity against known security features.

• Analyzes holograms, microprinting, and machine-readable zones (MRZ)
• Cross-references extracted data against issuing authority databases
• Detects pixel-level manipulation indicative of digital forgery
• Validates document consistency: checks that fonts, layouts, and data fields match the expected template for the document type and issuing country

A biometric authentication safeguard that distinguishes a live human presenter from a spoofing artifact, such as a photograph, video mask, or deepfake, during an identity verification session.

• Active liveness: prompts the user to perform random actions like blinking, smiling, or turning their head
• Passive liveness: analyzes micro-textures, light reflections, and blood flow patterns without user interaction
• Defeats presentation attacks including high-resolution prints, 3D masks, and screen replays
• Critical for remote onboarding flows where physical supervision is absent

The algorithmic process of comparing a live-captured biometric sample—typically a selfie—against the portrait image extracted from a trusted identity document to confirm they depict the same individual.

• Uses deep convolutional neural networks to generate facial embeddings
• Measures cosine similarity between the document portrait vector and the live capture vector
• Thresholds are tuned to balance false acceptance rate (FAR) and false rejection rate (FRR)
• Must account for aging, facial hair changes, and variable lighting conditions

The process of aggregating and evaluating an identity's publicly available online presence to assess its authenticity and longevity. A fabricated synthetic identity will lack the deep, time-consistent digital history of a real person.

• Checks for social media profiles with consistent biographical details and multi-year activity
• Analyzes email address age, domain registration dates, and breach history
• Evaluates phone number carrier records and tenure with mobile network operators
• Flags identities with no digital footprint or profiles created in a tight cluster of dates

A passive identification technique that collects unique attributes of a remote computing device to generate a persistent, probabilistically unique identifier. This detects repeat attackers even when they change identity credentials.

• Aggregates browser configuration, installed fonts, screen resolution, and WebGL rendering signatures
• Canvas fingerprinting exploits subtle GPU and driver differences in HTML5 Canvas rendering
• Detects emulators, virtual machines, and rooted devices indicative of fraud farms
• Links multiple application attempts from the same physical device despite IP rotation or cookie clearing

A verification method that generates dynamic, out-of-wallet questions derived from non-public data sources such as credit bureau files, property records, and vehicle registrations.

• Presents multiple-choice questions that only the legitimate identity holder should be able to answer
• Includes red-herring questions with plausible but incorrect answers to defeat guessers
• Measures response time and mouse movement patterns to detect scripted or researched answers
• Increasingly supplemented by biometric methods due to data breach proliferation