As financial crime continues to increase, banks are working to improve their risk-based anti-money laundering (AML) strategies. Because of the extensive data required for AML regulatory compliance and the growing intricacy of criminal methods, banks are under pressure to continually seek innovative tools to fulfil their regulatory responsibilities.
Considering this, artificial intelligence (AI) has become increasingly pivotal in the realm of AML compliance. According to Nasdaq’s 2024 Global Financial Crime Report, “as compliance pressure mounts and financial crime evolves, organizations that do not plan to increase spending on AI will need to identify alternative means to strengthen their financial crime management programs.” The report notes that “70% of respondents expect their organization to increase spending on AI or machine learning in the next 1-2 years.”
Globally, banks are realizing that AI tools can significantly improve their compliance effectiveness by identifying risks and criminal associations that may elude traditional manual and isolated compliance methods. Because of the limitations of current AML measures, coupled with the escalating complexity of threats and fast-expanding data sets for analysis, now is an opportune time to explore the potential of AI. AI offers not only greater risk identification and prevention capabilities, but also the scalability and adaptability required to counter the evolving threat landscape.
Promising AI/AML use cases
Leading banks are now pursuing promising AI use cases in the AML arena to deliver trusted outcomes responsibly. Some of these use cases include the following:
- Customer onboarding: Pattern recognition for biometric verification, optical character recognition, and natural language processing to validate documentation
- Risk assessment: Identification of direct and indirect risks through hidden connections
- Transaction monitoring: Anomaly detection to identify unknown typologies and hidden risks
- Alert investigation: Automated assessments and focused investigations
- Regulatory reporting: Automated population of suspicious transaction reports using natural language processing and natural language generation
- Model optimization: Automated tuning and optimization
- Case management: Intelligent case linking and grouping for holistic review and escalation
All the above use cases offer banks improved efficiencies, straight-through processing, explainability (i.e., how to take a machine learning model and explain the behavior in human terms), and auditability, which, in turn, improve their overall AML effectiveness.
Key challenges in integrating AI with AML
The introduction and use of AI and machine learning for AML compliance poses several risks that need to be carefully managed. Some of these risks include the following:
- Model complexity: The expanding size and intricacy of machine learning models underscore the need for effective risk management and control procedures.
- Explainability: The lack of explainability in machine learning models poses difficulties in easily validating, controlling, and governing their operations.
- Performance impacts: Machine performance may face challenges in situations where prior intelligence is lacking, and human experience, knowledge, and judgment are crucial.
- Staff training gaps: Insufficient training may hinder the ability of staff to use an AI-driven system, comprehend its functionalities, and address associated risks effectively.
- Data quality and algorithm concerns: Challenges related to data quality and biased algorithms can result in unintended outcomes, inaccurate predictions, and suboptimal decision-making.
Moving forward with AI-driven AML
How can banks best navigate the challenges of integrating AI and AML? Here are four key recommendations:
- Use AI to enhance, not replace, human intelligence: Despite the potential benefits of integrating AI and AML, there are debates about AI’s efficacy and the degree to which it should be trusted or even replace human analysis and decision-making. While reservations about widespread AI adoption are understandable, it's worth noting that the human brain is arguably the most complex and unpredictable system known. There is a prevailing belief that synergizing human insight and processes with AI can yield innovative work methodologies and superior outcomes, surpassing the effectiveness of deploying either humans or AI in isolation.
- Invest in continuous training: To fully explore and harness the potential of AI, we encourage banks to continuously enhance their understanding of AI’s capabilities, risks, and limitations through training. Ensure your staff on both the business and technology sides of your organization have a fundamental awareness of AI's impact on the business, its different uses, and the requirements for successful integration and outcomes.
- Establish a responsible use of AI framework: A robust and ethical framework governing the development and use of AI is crucial. Such a framework provides the necessary guardrails for ensuring emerging AI use cases and models are effective and, ultimately, deliver trusted outcomes. To learn more, check out these blogs from my colleague Dr. Diane Gutiw: Embracing responsible AI in the move from automation to creation and Guardrails for data protection in the age of GenAI.
- Find an AI partner. AI is complex and evolving. Collaborating with an experienced AI partner can provide access to the expertise required to keep pace. Look for a partner that can deliver not only innovative AI strategies, services and solutions, but also relevant industry know-how that integrates scientific rigor and a human-centric approach into the design and implementation of AI solutions.
CGI partners with banks across the globe to help them leverage the power of AI in different business areas, including AML. We combine our AML experience and intellectual property solutions (e.g., CGI Hotscan360) with AI-driven strategies, design and build services, operations and managed services, and intelligent solutions.
To learn more about our work, feel free to contact us.
