Fraud LensCopy Anchor Link

Fraud Lens is Prembly's centralized fraud intelligence engine and proactive defense network. Rather than forcing organizations to fight financial crime in isolation, Fraud Lens connects businesses to a shared consortium—the Fraud Lens—where verified bad actors, compromised accounts, and synthetic identities are cataloged with forensic evidence.

When a fraud ring or bad actor attacks one organization in the Prembly ecosystem, the entire network is instantly protected before that same entity can transact elsewhere.

1. What Businesses Stand to GainCopy Anchor Link

Traditional fraud defenses only react after an attack impacts your balance sheet. Fraud Lens transforms security into a pre-emptive, collective shield. By tapping into cross-institutional Network Overlap, businesses detect bad actors at the door without adding friction for legitimate customers.

Industry-Specific ImpactCopy Anchor Link

1. Fintechs & NeobanksCopy Anchor Link

• The Threat: Rapid, automated account creation, wallet draining, account takeover (ATO), and mule account creation used to launder stolen funds.
• What You Gain:
  • Pre-Emptive Sign-Up Shielding: Stops fraudsters from hopping between digital wallets and challenger banks. If a user is flagged for card draining at one fintech, their identity markers are blocked across all participating neobanks.
  • Reduced Manual Review Overhead: High-confidence matches (>= 0.95) trigger automatic blocks or step-up authentication, reducing operations queue volume by up to 80%.

2. Digital Lending & MicrocreditCopy Anchor Link

• The Threat: Loan stacking (borrowers simultaneously applying for loans across 5+ platforms within hours), synthetic identities created from stolen registry data, and intentional default syndicates.
• What You Gain:
  • Elimination of Loan Stacking: Instantly flags applicants who have defaulted or executed simultaneous loan runs across peer lenders.
  • Synthetic Identity Detection: Uses 1:N facial biometric vector scanning to catch individuals reusing the same face across different names or national identity numbers (e.g., NIN, BVN).
  • Direct NPL Reduction: Prevents non-performing loans (NPLs) and balance-sheet write-offs before capital is disbursed.

3. E-Commerce & MarketplacesCopy Anchor Link

• The Threat: Card-Not-Present (CNP) fraud, triangulation schemes, and "friendly fraud" (cardholders dishonestly filing chargeback disputes after receiving goods).
• What You Gain:
  • Chargeback Ratio Protection: Identifies serial dispute abusers and compromised credit cards before checkout, keeping dispute ratios safely below Visa and Mastercard monitoring thresholds (VAMP/MATCH).
  • Merchant Imposter Screening: Vets incoming marketplace sellers against known fraudulent corporate entities and bad-actor storefronts.

4. Traditional Banking & Financial InstitutionsCopy Anchor Link

• The Threat: Organized financial crime syndicates, stolen bank credentials, unauthorized wire diversions, and regulatory non-compliance fines.
• What You Gain:
  • Inter-Bank Threat Intelligence: Uncovers cross-institutional money laundering and coordinated syndicate activity before illicit funds transit downstream.
  • Privacy-Preserving Collaboration: Organizations share actionable threat signals and verified bad-actor hashes without exposing customer PII to competitors, ensuring full compliance with regional data protection mandates (GDPR, NDPR).
  • Regulatory & Audit Assurance: Every match includes a cryptographically verifiable audit trail suitable for regulatory filings and law enforcement dossiers.

2. Core Threat ClassificationsCopy Anchor Link

Fraud Lens categorizes threat vectors into 11 distinct operational types to enable granular rule triggers:

3. Detection & Screening ModalitiesCopy Anchor Link

Organizations can interrogate the Fraud Lens through three primary modalities:

1. Identity & Attribute ScanningCopy Anchor Link

Query individual identity markers—such as National Identity Numbers (BVN, NIN, SSN), bank account numbers, phone numbers, or email addresses—directly via REST API or the dashboard.

POST https://api.prembly.com/api/v1/fraud/portal/id-scan/
X-API-Key: <your_api_key>
Content-Type: application/json

{
  "id_type": "bvn",
  "id_number": "12345678901",
  "fraud_category": "bank",
  "verified_only": true
}

2. Facial Vector Biometric Scanning (1:N Matching)Copy Anchor Link

Fraud rings often alternate names and identity numbers while using the same physical photograph. Fraud Lens converts applicant selfies into high-dimensional facial embeddings and performs 1:N similarity searches across the entire Fraud Lens repository.

• Similarity Score >= 0.95 (Critical Match): Definitive match with a verified bad actor. Automatically flags or terminates processing.
• Similarity Score 0.70 – 0.94 (Strong Match): Elevated probability of synthetic manipulation or family impersonation. Routes into Case Management for operator review.

Face ScanCopy Anchor Link

Face comparison asks whether two images match. Face Scan asks a broader question: has this face been seen before, and in what company?

POST https://api.prembly.com/api/v1/fraud/face-scan/
X-API-Key: <your_api_key>
Content-Type: multipart/form-data

The scan runs three layers of analysis against a single submitted face:

1. Multiple ID Detection — checks whether the same face appears across several national identity records (BVN, NIN, Passport). One face bound to several distinct identities is a well-documented identity-farming pattern.
2. Fraud Bank Matching — matches the face against verified fraud reports, returning fraud categories, amounts, and report detail for both confirmed cases and unverified submissions.
3. Identity Consistency Analysis — scores overall identity confidence and raises security flags where information conflicts across sources.

The response carries an overall risk score from 0–100, a risk level of LOW, MEDIUM, HIGH, or CRITICAL, and an analysis_summary naming the specific pattern detected — for example, "Face matched to 3 different IDs".

For the biometric channels that compare and bind faces to identity records, rather than screening them for fraud, see the Biometrics guide.

3. Bulk Historical AuditsCopy Anchor Link

Audit existing customer portfolios, historical transaction batches, or merchant backlogs without hitting rate limits by uploading structured spreadsheets (.xlsx).

1. Format Spreadsheet: Include columns for identity parameters (id_type, id_number) or facial image URLs (image_url).
2. Dispatch Scan: Upload via the dashboard or submit programmatically to POST /api/v1/fraud/bulk-scan/upload/.
3. Download Results: Receive an enriched export containing row-level result_risk_score, result_risk_level, and result_fraud_detected flags.

4. Contributing Reports to the Fraud LensCopy Anchor Link

Organizations actively strengthen the consortium by submitting verified fraud cases:

Submission WorkflowCopy Anchor Link

1. Suspect Details: Input legal names, date of birth, state/region, and known identifiers (BVN, NIN, national ID).
2. Financial & Fraud Classification: Select the fraud category (e.g., Bank, E-Commerce, Telecom, Insurance), attack type, financial institution involved, account number, and monetary amount at risk.
3. Forensic Evidence: Attach supporting documentation (PDF dispute forms, transaction slips, chat transcripts, or altered identity images).
4. Consent & Attestation: Formal attestation that the submitted data represents verified fraudulent activity.

Review LifecycleCopy Anchor Link

Submitted cases undergo rigorous validation to prevent false accusations or retaliatory reporting:

• Pending Review: Newly submitted report entering verification queues.
• Under Review: Evidence is audited against third-party records and supporting proof.
• Approved: Case is confirmed, and the threat signatures are propagated across the global Fraud Lens shield.
• Declined: The submission lacked conclusive forensic evidence and was rejected.

5. Ecosystem IntegrationCopy Anchor Link

Fraud Lens functions as an integrated risk signal across the Prembly suite:

• Customer 360: Fraud hits automatically populate the customer's risk profile under the dedicated Fraud Tab, updating their composite risk score and alerting relationship managers.
• Workflows: Embed Fraud Scan nodes directly into automated onboarding pipelines to terminate suspect applicants before completing identity enrollment.
• Case Management: Anomaly matches and borderline similarity scores automatically spawn collaborative investigation cases for compliance and legal teams.
• Automated Webhooks: Receive immediate notifications (fraud.match_detected, fraud.report_approved) whenever an applicant or monitored account triggers a network hit.