Outbound Check Fraud (OBC), and True-Party Intention (TPI) Deposit Abuse Qorus-Infosys Finacle Banking Innovation Awards 2025

CONCEPT AND OBJECTIVE: AI is a powerful tool in the fight against financial fraud in consumer banking. By analyzing vast amounts of data in real-time, AI algorithms can identify suspicious patterns and prevent fraudulent activities. The objective of this innovative solution is to protect TD customers and the bank through AI-powered real-time transaction monitoring, behavioral biometrics, network analysis, predictive analytics and enhanced customer authentication.

Both models are AI systems developed to detect fraud at different points of contact with the customer. The OBC model aims to flag fraudulent out-bound checks for internal fraud agents to examine for further action to directly protect customer interests. The TPI model helps identify customers who are more likely to commit true-party intentional fraud, protecting the Bank from malicious users. Both models use a set of explanatory variables at the transaction level, as well as customer, account, and/or check data. Additional trend and domain knowledge-based features build the profile of the samples, allowing the models to predict the likelihood of fraud.

IMPACT: Fraud is a space where bad actors are always changing their techniques and avenues of attack. The Bank must stay vigilant to adapt to new threats which includes evolving our fraud AI models. The OBC model has evolved over time by updating its training set with new data as more is collected. With the expansion of available data over time, the performance of the OBC model has progressively improved, capturing a greater portion of fraudulent checks. Capacity to manually review checks for fraud is highly constrained at the scale on which TD operates. Our OBC model provided a breakthrough in performance with the ability to prioritize the most suspicious outbound checks for manual review. As one component in a sophisticated pipeline, this model has improved the proportion of fraudulent checks that are redirected to manual review, preventing losses that would be imposed on the Bank through fraudulent checks leaving the institution. In a similar vein, TD needs to automate most check clearance and deposit holds via business strategies. The TPI model has complemented existing processes by flagging atypical customer activity and routing them for manual review. This allows TD to provide streamlined customer service for the vast majority of customers who act in good faith, for example by reducing or eliminating deposit hold times, while protecting the Bank from intentional fraud.