As digital wallets grow in adoption for everyday transactions, they’ve also become prime targets for fraudsters. Suspicious top-ups, unauthorized withdrawals, and account takeovers often hide within normal transaction patterns — making them hard to catch.

Solving this problem means fewer chargebacks, less fraud loss, and better protection for both users and institutions.

The Challenge:

• Traditional systems struggle to detect low-frequency, high-risk behaviors

• Fraud evolves quickly, making static rules obsolete

• False positives frustrate users and drain resources

• Multi-device, multi-location behaviors complicate detection

Quantum-Inspired Solution

This use case focuses on how quantum-inspired models can detect subtle and emerging threats in real-time, enhancing trust in financial systems and improving user safety.

Behavioral Graph Modeling

HessQ maps transactions, devices, locations, and behaviors into a multi-dimensional graph, allowing detection of unusual clusters

Optimizes scoring to identify outliers based on context (e.g., sudden spikes in use, new device pairs, or unusual time of use)

Learns and adjusts to new behavior patterns without retraining — perfect for fast-moving fraud scenarios

QUBO-Based Anomaly Scoring

Real-Time Adaptability

The Results:

• +45% increase in anomaly detection accuracy

• 60% reduction in false alerts

• From 3 hours → 2 minutes to flag fraud attempts

• Improved trust and lower churn for wallet users