Fingerprint has added AI-powered recommendations to its Suspect Score product, shifting fraud detection from static rule-based scoring to an adaptive model trained on each customer’s own labeled fraud data, as part of a broader platform strategy targeting fraud patterns that evolve faster than manual tuning can address.
The enhanced Suspect Score uses machine learning to analyze a customer’s historical fraud and legitimate traffic patterns alongside Fingerprint’s Smart Signals, a suite of device intelligence indicators, to generate automatically optimized signal weights customized to that customer’s specific fraud profile. Rather than applying a fixed universal scoring model, the system adjusts which signals carry the most weight based on observed patterns in the customer’s own data, reducing false positives that result from mismatched generic weightings. Users can preview all recommended signal weight changes before applying them and can retrain their scoring model on demand as threat patterns shift.
The feature is immediately available to all Fingerprint customers with access to Smart Signals, who can begin training customized models through the Fingerprint dashboard without additional setup. Fingerprint, which reported 175 percent revenue growth driven by AI fraud threats in April 2025, reported 65 percent ARR growth and surpassed one billion device identifications per month as of February 2026.
“Fraud patterns vary by business and evolve constantly, rendering manual tuning obsolete. Our AI-powered recommendations remove that bottleneck by training on each customer’s labeled data, making Suspect Score customizable, accurate, and easy for customers to use,” said Valentin Vasilyev, CTO and co-founder of Fingerprint.
Sumsub embedded Fingerprint’s device intelligence last year to strengthen its fraud prevention platform, making the Suspect Score’s AI-trained signal weighting directly relevant to identity verification providers using Fingerprint as a fraud layer. The adaptive Suspect Score follows Fingerprint’s launch of AI agent detection in February to distinguish legitimate automation from malicious bot traffic, continuing a pattern of applying machine learning to increasingly complex fraud scenarios across the platform.
Fingerprint’s technology operates as device intelligence using browser fingerprinting and behavioral signal analysis rather than biometric capture, positioning the Suspect Score as a fraud prevention layer within identity verification and onboarding workflows rather than a standalone authentication mechanism.
Sources: Fingerprint, Business Wire, The Paypers
–
By the ID Tech Editorial Team
