Generative AI and financial crimes: a quantitative systematic literature review

Abstract Purpose The proliferation of large language models (LLMs) and generative artificial intelligence (GenAI) applications has provided ample opportunities for crime, including technology-facilitated financial crimes. The present study conducted a quantitative systematic literature review to examine the evolving intersection of GenAI and financial crime. Specifically, the study identified keyword concentrations, latent research topics, and thematic relationships within this emerging domain to explore how current scholarship understands the role of GenAI in financial crimes. Methods Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020, this systematic literature review employed three quantitative analytical methods—bibliometric analysis, topic modelling, and knowledge graph analysis—to reveal trends, concentrations, and connections of prominent keywords and topics that emerged from the extant literature. Results With the assistance of the PRISMA 2020 framework, a total of 94 studies were incorporated for the quantitative systematic review. The bibliometric analysis identified five keyword clusters, while the topic modelling and knowledge graph revealed six latent research topics with nuanced patterns, highlighting a growing concentration on automated financial crimes that are distinctive from human-centric social engineering. The results also revealed the dual-use nature of GenAI in both facilitating and preventing financial crimes. On one hand, GenAI has been widely misused in financial cybercrimes such as algorithmic fraud, deepfake attacks, and smart contract exploitation. On the other hand, GenAI has enhanced crime prevention capacities, such as detection capabilities and vulnerability screening. Conclusions While GenAI facilitates various criminal opportunities for financial crime, it also provides insight into effective crime prevention strategies. This study demonstrated that research is increasingly focused on the technical and adversarial dimensions of the dual-use nature of GenAI, outlining a structural distinction between human-centric social engineering and automated financial crimes. The findings shed light on the importance of recognising the evolutionary landscape of financial crimes enabled by GenAI and the significance of embracing forward-looking governance frameworks for regulatory compliance in decentralised financial (DeFi) systems.