Fingerprint Launches Fraud Score Predictive Tool
The tool builds organization-specific fraud models from each client's own outcome data, combining device intelligence with machine learning.
Updated on Nov. 11, 2026 in Financial Crime
Fingerprint has rolled out Fraud Score, a predictive fraud tool that merges its device signals with outcome data supplied by each client. In one head-to-head test, Fraud Score flagged almost 50% more fraud than the client's in-house system.
Why it matters
The improvement in fraud capture occurred without any rise in legitimate users being blocked, a trade-off businesses often struggle to achieve. Fraud Score's models are retrained on an ongoing basis as customers submit fresh outcomes.
Fraud Score runs dedicated models for payment fraud, account takeover, multi-accounting, fake account creation, promotion abuse, platform abuse and content abuse. Predictions can be used directly or fed into existing decision engines through Fingerprint's API responses.
The players
Fingerprint
Fingerprint is a company that provides device intelligence and identification technology used by businesses to detect and prevent online fraud.
The details
Fraud Score merges Fingerprint's device signals with client-supplied outcome data to produce organization-specific models. Fingerprint retrains the models on an ongoing basis as customers submit fresh outcomes.
Timeline
November 11, 2026: Fingerprint publishes the announcement of the Fraud Score rollout.
Tech Race
Fraud Score extends device-intelligence fraud detection by shifting from generic, shared fraud models toward models trained on each client's own outcome data. It positions Fingerprint's device-signal approach against rival fraud-scoring tools that rely on broader, non-customized training.
Businesses using Fingerprint can route Fraud Score predictions directly into their decision engines via API responses. Legitimate customers may see fewer false blocks if the model's improved precision holds across implementations.
The takeaway
Organizations whose fraud models are trained only on generic data may be missing a meaningful share of fraud that per-client outcome training can catch. Reviewing whether existing fraud tools learn from your own outcomes is a practical first step.
Further reading
For more on how fraud schemes are tracked and prosecuted, explore Financial Crime.
