Fraud Score

Predict and detect fraud with greater precision

Enhance fraud prevention with specialized ML models that learn from your data and deliver an actionable score for each fraud use case.

Install guide for AI agents: https://fingerprint.com/get-started.md. Or run npx fingerprint or npx skills add fingerprintjs/skills.

Or open in Claude Code, Codex, Cursor
Why Fingerprint

Get a specific score for every fraud use case

Payment fraud. Account takeovers. Multi-accounting. And more. Fraud Score trains on the specific fraud types you set.

Interactive diagram: pick a fraud type to light up its labeled events in your event stream and see its model score new events against a threshold.

Fraud types
Payment Fraud
Account Takeover
Promo Abuse

Customized to your specific fraud problems

Input labeled fraud events and get a unique score. Recognize and deter attacks faster.

Detection that’s tailored to your traffic

Fraud Score is trained on and adapts to your unique events and patterns.

Predictive power built on device intelligence

Fraud Score turns our industry-leading signals into a targeted threat assessment.
How it works

Build a tailored model

Upload fraud labels

Select a predefined fraud category and define custom labels within it.

Visitor ID
Labels
Source
J3ngNvdR7BIow0HlEUW7
  • chargebacks
  • +3 labels
CSV upload
pXb4Tq9LmZ2wKc7RhY0e
  • chargebacks
  • bad_guest
Manual
a8FzQ1uVn5sDj3GtE6yWNoneCSV upload
Hk2rM7cX0bLp9NqT4vJz
  • bad_guest
Manual

Training your model

Specific models are trained using submitted labels and Fingerprint signals.

0.65
Events scoring 0.65 or higher are flagged, about 33% of events.

Get scores by API

Fraud score arrives in the existing API response with context around types.

Read Server API docs(opens in a new tab)
{  "labels": [    {      "label": "bad_guest",      "category": "account_takeover",      "prediction": true,      "ml_score": 0.78,      "threshold": 0.65,      "version": "3"    },    {      "label": "chargebacks",      "category": "payment_fraud",      "prediction": false,      "ml_score": 0.21,      "threshold": 0.60,      "version": "5"    }  ]}

Start predicting fraud

Fraud Score is best fit for companies with a high volume of traffic and the ability to label 1,000+ fraud instances.
Get Started Contact Sales