Skip to main content
AntifraudAI is a real-time fraud detection platform that helps businesses stop fraudulent transactions and suspicious user activity before they cause damage. By combining machine learning models, behavioral signals, and configurable rule engines, AntifraudAI screens every event in milliseconds — returning a clear risk score and actionable decision so your systems can respond instantly, without slowing down legitimate customers.

​
What AntifraudAI Does

Every time a transaction or user event is submitted to the AntifraudAI API, the platform runs it through a multi-layered analysis pipeline:
  1. Real-time screening — Each request is evaluated the moment it arrives, with a median response time under 50ms so your checkout or login flow is never blocked.
  2. Risk scoring — AntifraudAI returns a numeric risk score between 0 (no risk) and 100 (highest risk), giving you fine-grained control over how your application responds.
  3. Decisions — Alongside the score, every response includes a plain-language decision:
    • allow — the event appears legitimate; proceed normally.
    • review — the event has elevated risk signals; flag it for manual or automated review.
    • block — the event is highly likely to be fraudulent; reject it.
  4. Signals — A signals array explains why a score was assigned (e.g., high_velocity, new_device, ip_mismatch), giving your team full visibility into every decision.

​
Key Features

Real-Time Scoring

Screen transactions and user events in under 50ms with ML-powered risk scores from 0 to 100, so your customers never notice the check.

Custom Rules

Build and deploy your own rule logic on top of the ML baseline — block by country, velocity, device fingerprint, or any custom attribute you send.

Case Management

Escalate review decisions into a built-in case queue where your fraud analysts can investigate, annotate, and resolve events from a single dashboard.

Webhooks

Receive instant push notifications when a decision changes — ideal for async review flows where a human or secondary model re-scores an event.

Sandbox Testing

A fully isolated sandbox environment mirrors production behavior so you can build, test, and break things safely before going live.

Event Signals

Enrich every request with behavioral and device signals — IP geolocation, device fingerprint, session velocity, and more — to increase scoring accuracy.

​
How the API Works

Integrating AntifraudAI follows a simple request/response pattern. Your server sends a POST request describing the transaction or event, and AntifraudAI responds with a risk score, a decision, and the signals that influenced the outcome — all in a single round-trip. Request: Send event data as a JSON body with your API key in the Authorization header. Response: Receive a structured JSON object you can act on immediately inside your own business logic. Here’s the simplest possible example — screening a payment transaction:
Response:
In this response, a risk_score of 72 triggered a review decision, and the signals array tells you exactly what drove the score — in this case, an unusually high transaction velocity for the user combined with a device that hasn’t been seen before.

​
Next Steps

Quickstart

Make your first API call and receive a fraud decision in under 5 minutes.

Authentication

Learn how to create API keys and authenticate every request securely.

How It Works

Dive deeper into the scoring pipeline, ML models, and decision logic.

API Reference

Explore every endpoint, parameter, and response schema in full detail.