Artificial intelligence and machine learning applications in underwriting, fraud detection, and risk assessment

The discipline of underwriting is generally fairly conservative, but because insurers are trying to squeeze more information and value out of increasingly abundant data sources, increasing interest has been applied to the use of Artificial Intelligence and Machine Learning technologies in underwriting. These discussions are primarily concentrated in Personal Lines where the large volume of homogeneous data can enable AI and ML to produce results rapidly, but as more data is collected in Small and Medium-sized Enterprises, Delegated Authority Schemes, and Commercial businesses, this technology will inevitably be applied more widely. Insurers have recognized that they are sitting on a vast data mountain, leveraging them through the use of AI and ML technologies to squeeze the last ounce of value and insight from this data. These applications run the whole gambit from the direct use of AI and ML within underwriting models to the deployment of AI-driven tools that assist underwriters in their operations. The workflow across personal, SME, and commercial underwriting is changing based on this technology adoption. AI is complementing and augmenting underwriting decisions through predictive scoring and new insights distilled or sourced via AI (Rahman et al., 2023; Kim et al., 2024; Patel et al., 2024).

While the use of AI and ML in risk selection is the original driver for its adoption, the technology is impacting other areas of underwriting workflow. Be it through automating key reports or general tasking, collating information across sources and enabling ratemaking to be completed rapidly, improving the general experience and workflow users experience, or enabling the support of less sophisticated users through chatting interfaces. AI and ML are having a sustained impact on the underwriting domain. Risk selection, file assessment, business classification, portfolio review, business optimization, and service or partner selection are only a small selection of the many applications across the underwriting discipline (Singh et al., 2023; Silva et al., 2025).