Unlocking the Value of GenAI in Financial Crime Compliance

Generative AI (GenAI) has swiftly evolved from a buzzword to a pivotal force in financial services, promising transformative opportunities while introducing new risks. With its potential to revolutionize banking operations, enhance financial crime detection, and streamline compliance, the conversation has shifted from whether financial institutions will leverage it to how they will do so effectively.

Exploring Cost and Efficiency Savings

During a recent panel discussion in Singapore, hosted by Regulation Asia in collaboration with SymphonyAI, industry leaders deliberated on the ways GenAI can enhance financial crime compliance programs. Craig Robertson, Financial Crime SME at SymphonyAI’s APAC Financial Services unit, emphasized that most firms are striving to improve automation, become more data-driven, and achieve greater efficiency and cost savings.

He noted that AI applications in transaction analysis, risk detection, external web searches, and KYC processes align with this drive. “For most firms, it’s about thinking through the cost and efficiency opportunities to determine which AI use cases to pursue first.”

DBS Bank, an early adopter of AI in financial crime compliance, has focused on improving efficiency over the past seven years. According to Lam Chee Kin, Managing Director & Head of Group Legal & Compliance at DBS, efficiency is crucial due to the finite nature of resources. “If you don’t also focus on efficiency, there is a natural limit to what you can do, and things will not be sustainable,” he explained.

A well-structured technology stack plays a vital role in this journey. It took DBS five years to establish an effective data architecture, computing suite, workflow, and reporting dashboard, enabling them to self-fund numerous financial crime initiatives.

Key GenAI Use Cases in Financial Crime Compliance

One of the most significant applications of GenAI in financial crime compliance has been the automation of suspicious transaction reports (STRs). DBS has utilized GenAI to generate first drafts of STRs, ensuring consistency and structuring key KYC and case information effectively. While human oversight remains necessary before submission, the technology has streamlined the process and eliminated inconsistencies in writing.

Bank of Singapore (BOS) has also leveraged GenAI for STR drafting and financial crime risk detection, particularly in adverse media analysis. Ng Kok Keong, Head of FCC Intelligence & Innovation at BOS, highlighted the integration of NLP for entity extraction and resolution within network analysis. “GenAI helps to generate additional insights from adverse news quickly, enhancing our risk scoring and understanding of unusual patterns,” he said.

Improving Effectiveness in Risk Detection

Beyond efficiency gains, GenAI significantly enhances risk detection capabilities. Robertson emphasized that AI not only “augments the human experience” but also allows professionals to focus on making critical risk assessments. Ng noted that AI provides a more holistic approach to risk assessment, moving beyond traditional rules-based systems that often generate high volumes of false positives.

By leveraging contextual monitoring, financial institutions can develop a more sophisticated understanding of client behavior, allowing them to better distinguish legitimate transactions from potential fraudulent activity. According to Lam, the agility to respond to rapidly evolving fraud tactics is crucial for improving overall effectiveness.

Ecosystem Benefits and Data Sharing Challenges

A critical component of GenAI’s success in financial crime compliance is its ability to leverage ecosystem benefits. Initiatives like Singapore’s COSMIC platform, which facilitates financial crime risk information sharing among major banks, exemplify this approach. However, data sharing challenges remain, as different financial institutions lack a standardized method for storing and harmonizing data.

To address this, Lam suggests focusing on label sharing instead of exchanging raw data, as this minimizes privacy concerns and cybersecurity risks. Additionally, collaboration across industries—including digital payments, telecom, e-commerce, and social media—could enhance fraud detection efforts in digital scams.

A New Era of AI-Driven Investigations

As AI adoption accelerates, financial institutions must ensure transparency and explainability in their AI models. Ng emphasized the need for investigators to be trained in AI methodologies, ensuring they can interpret AI-driven insights effectively.

BOS is exploring Model Ensembling—combining GenAI with existing AI models—to maintain explainability while enhancing fraud detection. “The clarity on model features highlighting risk typologies helps guide data analytics-trained investigators in uncovering fraud and financial crime risks more effectively,” Ng said. This effort is part of BOS’ “Investigations Reimagined” initiative, aimed at transforming traditional investigative practices with AI-driven insights.

Strategic Recommendations for AI Adoption

To support financial institutions in effectively integrating AI into their financial crime compliance programs, SymphonyAI has commissioned a report in collaboration with Regulation Asia. The report examines AI adoption maturity, implementation challenges, regulatory drivers, and strategic recommendations for successful AI deployment in financial services.

As financial institutions continue to unlock the value of GenAI, the focus must remain on balancing efficiency with effectiveness, ensuring transparency, and fostering industry-wide collaboration to combat financial crime more proactively and intelligently.