# HSBC Dynamic Risk Assessment AI Reduces AML False Positives by 60 Percent > The global bank partnered with Google Cloud to deploy an AI system that improves financial crime detection efficiency across its operations in multiple markets. *Published 2026-08-17 · By Diane Okafor* Dynamic Risk Assessment is an artificial intelligence system that HSBC co-developed with Google Cloud to identify financial crime in high volumes of banking transactions. ## Executive Summary HSBC is a major global bank operating in the financial services sector with extensive anti-money laundering responsibilities across numerous jurisdictions. The institution maintains operations that require continuous monitoring of transaction activity to comply with regulatory standards. The company implemented the Dynamic Risk Assessment AI system through a partnership with Google Cloud to address longstanding challenges in detection accuracy. The Dynamic Risk Assessment AI system was deployed to analyze transaction patterns and assess risk levels in real time. This approach replaced elements of traditional rules-based monitoring that had produced high volumes of alerts requiring manual review. The deployment focused on improving the precision of financial crime identification processes. This implementation resulted in a 60 percent reduction in false positive alerts and two to four times more detections of suspicious activity. The outcomes allow compliance teams to allocate resources more effectively toward higher priority cases. HSBC received the Celent Model Risk Manager of the Year 2023 award for the AML AI implementation. ## What background context led to the development of AI tools for financial crime at HSBC? HSBC processes between 980 million and 1.2 billion transactions each month that require screening for financial crime. This high volume creates substantial operational demands on compliance resources. The bank operates across 62 jurisdictions where regulatory expectations for transaction monitoring continue to evolve. Traditional rules-based systems often generated false positive rates in the range of 90 to 95 percent. These rates led to unnecessary reviews of legitimate customer activity. The resulting workload reduced the efficiency of financial crime prevention efforts. The bank sought advanced technology to improve the accuracy of its detection processes. This need arose from the combination of transaction scale and the limitations of static rule sets. The pursuit of AI solutions aimed to address these specific operational constraints. ## How did HSBC and Google Cloud collaborate on the Dynamic Risk Assessment system? HSBC partnered with Google Cloud to co-develop the AI system known internally as Dynamic Risk Assessment. The collaboration involved joint work on model design and integration with existing data infrastructure. This partnership enabled the incorporation of advanced analytics capabilities tailored to banking compliance needs. The development process included pilot testing that began in 2021. Initial production deployment occurred in the United Kingdom and Hong Kong. The co-development approach ensured alignment with HSBC's internal requirements for financial crime detection. The system uses artificial intelligence to assess risk in real time based on transaction patterns. Google Cloud provided the platform for building and scaling the solution. The partnership model supported the creation of a customized tool for AML purposes. ## What technical specifics define the AML AI implementation at HSBC? The AI solution analyzes transaction data to identify potential money laundering indicators more precisely than previous methods. It incorporates multiple variables including customer history and geographic factors into risk scoring. The model was trained on historical data to distinguish legitimate activity from suspicious cases. Deployment leverages Google Cloud infrastructure for scalability across global operations. The system operates by assigning risk scores to individual transactions. This allows for dynamic assessment rather than reliance on fixed thresholds. The implementation integrates with HSBC's existing transaction processing workflows. Machine learning components enable continuous improvement as new data is incorporated. The technical architecture supports analysis of the full monthly transaction volume. ## What measurable business outcomes resulted from the AI deployment? The system achieved a 60 percent reduction in the number of false positive cases. This reduction directly addresses the historical issue of unnecessary customer inquiries. Investigators now encounter two to four times more actual instances of suspicious activity. The shift improves the productivity of financial crime teams by focusing efforts on higher value alerts. Compliance resources can be redirected from low-yield reviews to more complex investigations. The overall effectiveness of the AML program increased as a result. ## How does the AI system performance compare to traditional methods? Performance comparison between traditional AML methods and HSBC's AI systemMetricRules-Based SystemDynamic Risk Assessment AIFalse Positive AlertsHigh volume at 90-95 percent rate60 percent reductionSuspicious Activity DetectionsBaseline level2-4 times increaseTransaction Volume Processed980 million to 1.2 billion monthlySame volume with improved accuracyDeployment MarketsInitial limited scopeUK and Hong Kong with expansion planned ## What implications does this have for stakeholders in the banking industry? Other financial institutions face similar challenges with false positive rates in compliance operations. Adoption of comparable AI systems can lead to cost savings in compliance staffing. Improved detection accuracy also supports better customer relations by reducing unnecessary contacts. Regulatory bodies may see benefits in more effective detection of financial crime across the sector. The approach demonstrates the potential for cloud-based AI partnerships in regulated industries. Banks can use such systems to meet regulatory requirements with greater efficiency. The quantified results provide a benchmark for peers evaluating similar investments. Stakeholder groups including compliance officers and risk managers can assess the operational impacts. The case illustrates how AI can transform aspects of enterprise compliance functions. ## What expert commentary has been provided on the AML AI results? > Historically, we had a high number of false positives, meaning that we were calling customers unnecessarily to ask them about what turned out to be completely legitimate activity. Now, we have 60% fewer false positive cases.Jennifer Calvery, Group Head of Financial Crime, HSBC The Celent award recognized the model risk management aspects of the implementation. This recognition highlights the robustness of the AI model in a compliance context. The award validates the approach taken in developing and deploying the system. Jennifer Calvery also stated that HSBC partnered with Google to co-develop the AI system used to check for financial crime. The system is known internally at HSBC as Dynamic Risk Assessment. This commentary underscores the collaborative nature of the project. ## What expansion plans exist for the Dynamic Risk Assessment system? HSBC plans to distribute the technology across additional markets beyond the initial deployments. The success in the UK and Hong Kong supports scaling to the full set of 62 jurisdictions. Further refinements to the model may occur as more data becomes available from expanded use. The expansion will involve adapting the system to local regulatory environments in new markets. Data from the initial deployments informs the strategy for broader rollout. The technology distribution aims to extend the performance gains achieved so far. ## What key takeaways should other enterprise leaders consider from this case? - Enterprise leaders should evaluate current false positive rates in compliance systems to identify specific improvement opportunities in AML processes. - Consider partnerships with technology providers to co-develop specialized AI solutions that align with internal requirements. - Pilot AI systems in key markets before broader rollout to validate performance gains in real operational conditions. - Measure outcomes in terms of both alert reduction and increased detection accuracy to assess full impact. - Seek recognition from industry analysts to validate the implementation approach and share learnings with peers. ## Sources 1. [HSBC reduced false positive cases by 60% and finds two to four times more financial crime with the AI system.](https://www.hsbc.com/news-and-views/views/hsbc-views/harnessing-the-power-of-ai-to-fight-financial-crime) 2. [AML AI identifies two to four times as much suspicious activity as the previous system, while reducing the number of alerts by 60%.](https://cloud.google.com/blog/topics/financial-services/how-hsbc-fights-money-launderers-with-artificial-intelligence) --- Source: https://aiintelreport.com/enterprise-ai/hsbc-dynamic-risk-assessment-aml-ai-win Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt