Fraud Detection
Every signal.One risk engine.Real-time defence.
Fraud isn't one check, it's a moving target. Myaza scores every customer and every action across four layers at once: real-time event monitoring, watchlist screening, device and signal intelligence, and a living trust score. Here's how it works, and how your team puts it to work.
工作原理
一个流程，端到端。
One engine that turns identity, behaviour, device and screening signals into a live, explainable risk decision.
- 1
Connect what you already have
Point verifications, transactions, logins and session events at one API. Nothing bespoke, the app signals you already collect become risk inputs, resolved to the right customer automatically.
- 2
Every signal folds into a trust score
Monitoring, screening, device intelligence and biometrics fold into a dimensional, explainable trust score per customer that decays over time and moves as their behaviour does.
- 3
Policies decide in real time
A no-code decision graph reads the score and the raw signals to allow, review, decline or challenge, synchronously, with no batch job and no engineering ticket to change it.
- 4
Your team investigates and reports
REVIEW and BLOCK decisions raise alerts that roll up into cases; reportable activity becomes a regulator-ready goAML SAR on a tamper-evident trail.
背后的支撑
底层的各个层次。
Real-time event monitoring
Every event, a transaction, login, session or action, is scored the moment it arrives against rules you tune per tenant and custom rules you author with no deploy. Transactions get the AML rules (velocity, structuring, geo, anomaly); logins and sessions get the account-takeover rules.
Sanctions, PEP & adverse media
Screen every customer at onboarding and re-screen on a risk-based schedule against sanctions, PEP and adverse-media lists across 200+ jurisdictions, sanctioned crypto wallet addresses included.
Device & signal intelligence
Device fingerprinting and IP analysis surface multi-accounting, emulators, datacenter IPs and velocity, plus account-takeover signals like a brand-new device or a country hop too fast to be travel.
A living trust score
Every signal folds into an explainable fraud-risk and identity-confidence score, with the top factors always shown. It decays over time and moves as the customer does, never a black box.
Decisions that act
A visual decision graph branches on the score and the raw signals to approve, review or decline, or automatically mint a step-up re-verification challenge instead of blocking a good customer.
Alerts, cases & SAR filing
Repeat firings roll into one alert, alerts group into cases with four-eyes controls, and reportable activity exports as goAML XML, the whole investigation on a tamper-evident audit trail.
Frequently asked
Fraud Detection常见问题。
KYC verifies who someone is at sign-up. Fraud detection keeps working afterward: scoring every transaction, login and action against a living risk profile, re-screening customers as watchlists change, and catching takeover and coordinated fraud that a one-time check can't see.