When it comes to the highly regulated world of banking, identifying and mitigating risks associated with money laundering is a critical challenge. A large US bank, serving approximately 40,000 corporate and institutional clients, sought to enhance its transaction monitoring processes to detect suspicious activities more efficiently. Partnering with Gradient, the bank looked to AI for a solution to help reduce transaction risk and increase the accuracy of detecting suspicious activity.
The bank faced a multifaceted problem in its anti-money laundering (AML) efforts. Traditional risk modeling techniques were inadequate due to their reliance on structured data, leaving vast amounts of unstructured data untapped. This limitation hindered the bank's ability to accurately identify potential money laundering activities, resulting in a high rate of false positives and inefficient use of resources. The primary challenge was to integrate this unstructured data into the risk assessment process, enhancing the detection capabilities without compromising operational efficiency.
The AI-Driven Solution
Gradient introduced a comprehensive, AI-powered solution to address these challenges, centered around Albatross, Gradient’s proprietary domain-specific LLM in finance. The solution can be broken down into two phases.
Phase 1: Enhanced Risk Scoring
By augmenting traditional risk scores with unstructured customer data and activity insights, Gradient's model was able to analyze a broader dataset, providing a more nuanced understanding of transactional risk.
Phase 2: AI-Powered AML Detection
Gradient's AI-powered solution transitioned the bank from a rule-based transaction monitoring to a fully AI-powered approach - enabling automated AML detection, triaging, and reporting. This approach allowed for the identification of complex patterns and anomalies that traditional systems might overlook. With this transition, the team was able to build out key features to further facilitate faster and more effective decision making including:
Providing AI-Powered Risk Scores: With an AI-powered approach, you can replace the manual process of having to define parameters that are generally associated with rule-based transaction monitoring. Using the bank’s extensive data assets, Gradient was able to further train and fine-tune Albatross to maximize it’s ability to identify suspicious, potential money-laundering activity and provide a comprehensive view of risk scores.
Ranking Potential Risks: By leveraging a comprehensive analysis of the Bank’s private data, the model automatically identifies the highest rated potential risks based on a weighted system. To do this, our custom model scrutinizes various data and uncovers patters, behaviors or anomalies critical to bank including: transactions, customer interactions, account details, company information, and more.
Simplifying Risk Score Interpretation: The AI-powered solution provide a breakdown of key indicators based on the analysis. This means the financial institution is able to understand the generated risk scores to help expedite the investigation process or further improving the taxonomy.
The Impact
The collaboration between our client and Gradient yielded significant benefits including:
Increased Risk Detection
The AI-powered solution enabled the bank to detect 3-4 times more instances of first and third-party fraud by automating the detection process, resulting in higher efficiency and accuracy.
Decrease in Operational Costs
By reducing false positives by more than 50%, the bank could reallocate resources to focus on high-risk, actionable insights, streamlining the investigation workflow.
Enhanced Governance
The AI-driven approach supported regulatory compliance and internal risk management by providing auditable and explainable outputs, complete with automated reporting capabilities.
Conclusion
The partnership between our client and Gradient marks a significant advancement in the use of AI within the commercial banking sector to combat money laundering. By leveraging AI to enhance transaction monitoring and risk assessment, the bank not only improved its compliance posture but also achieved greater operational efficiency and accuracy in detecting fraudulent activities.
