Fingerprint Unveils AI-Enhanced Suspect Score to Combat Evolving Fraud Schemes

New AI-powered recommendations enable enterprises to train fraud scoring on their own data for more accurate detection

CHICAGO--(BUSINESS WIRE)--Fingerprint, a leader in device intelligence for fraud prevention, today announced the addition of AI-powered recommendations to its Suspect Score solution. The enhancement gives Fingerprint customers an adaptive, intelligent fraud score trained on their own labeled data, improving detection accuracy while maintaining full transparency and control.

Static scoring models fail to keep pace with increasingly dynamic, traffic-specific fraud patterns. Fraud teams lack the time and resources to continuously analyze signal interactions and tune model weights for their unique needs. With Fingerprint's AI-powered recommendations, fraud teams can eliminate manual tuning, save valuable time and resources, and make their fraud detection adaptive to evolving threats.

"Fraud patterns vary by business and evolve constantly, rendering manual tuning obsolete," said Valentin Vasilyev, CTO and co-founder at Fingerprint. "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."

Adaptive Fraud Detection for Evolving Threats

Sophisticated AI agents and bots can bypass static detection models, leaving organizations vulnerable to modern fraud tactics. Compounding the challenge, legitimate users are increasingly adopting privacy tools like VPNs, complicating traditional signal weighting.

Fingerprint's enhanced Suspect Score addresses this issue with a production-ready machine learning (ML) system. Built on Fingerprint's suite of Smart Signals — actionable real-time device intelligence insights — Suspect Score already delivers powerful fraud indicators. Now, enterprise fraud and security teams can upload labeled fraud data to train the ML system on their unique traffic patterns as threats evolve.

Using this data, Fingerprint's updated Suspect Score:

• Intelligently analyzes customer data alongside Smart Signals to generate optimized signal weights tailored to a customer’s specific fraud patterns
• Adjusts signal weights based on patterns observed in a customer’s fraud data to reduce false positives while maintaining accuracy
• Provides a preview of all recommendations before customers apply changes with a single click, giving users full visibility and control over their scoring

As threats evolve, organizations can retrain their scoring with up-to-date data to keep detection aligned with real-world fraud behavior.

A New Standard for Fraud Detection

Fingerprint's AI-powered Suspect Score recommendations shift fraud detection from static to adaptive. By tailoring detection to each organization's unique traffic patterns, Fingerprint sets a new data-driven standard, providing continuous optimization without sacrificing transparency or control.

AI-powered Suspect Score recommendations are now available to all Fingerprint customers with access to Smart Signals. Existing customers can begin training customized scoring models through the Fingerprint dashboard.

About Fingerprint

Fingerprint detects the intent of human and agentic visitors. Its device intelligence platform identifies over 1 billion unique devices every month and processes hundreds of signals to help fraud teams distinguish trusted visitors from bad actors at speed and scale. Over 6,000 companies, including innovators like Dropbox, Booking.com, and checkout.com, use Fingerprint every day to recognize high-risk activity in real time, prevent fraud attacks, and deliver frictionless user experiences. Learn more at fingerprint.com.

Frequently Asked Questions

How can enterprise fraud teams customize fraud scoring to their own traffic patterns?

Static fraud scoring models can't keep pace with evolving, business-specific fraud tactics. Fingerprint's Suspect Score intelligently analyzes and trains on each organization's own labeled data to generate signal weights tailored to their specific traffic — reducing false positives and improving detection without manual tuning. It's built on over 100 real-time device intelligence signals, and teams can update the tuning by uploading up-to-date data as threats change.

How does machine learning improve fraud detection accuracy without sacrificing transparency?

Machine learning improves fraud detection accuracy by training on an enterprise's labeled fraud data to tune weights based on traffic-specific patterns. Fingerprint's Suspect Score surfaces all AI-generated recommendations before they are implemented, giving fraud teams full control and visibility into how their scoring works. Organizations get adaptive detection without giving up oversight.

How can fraud teams reduce false positives without increasing fraud exposure?

As more legitimate users adopt privacy tools like VPNs, traditional signal weighting increasingly catches the wrong people. Fingerprint's Suspect Score analyzes a company's own fraud data to optimize individual signal weights — for example, by lowering the weight of a VPN signal where the data supports doing so. The model can be retrained as new patterns emerge, keeping detection calibrated without the need for continuous manual intervention.