Fighting Financial Crime with AI.

This article focuses on KPMG's innovative use of artificial intelligence (AI), specifically natural language processing (NLP) and large language models like GPT-3, to combat money laundering by accurately predicting the nature of businesses. KPMG's solution analyzes publicly available company information, such as websites, to assign Standard Industrial Classification (SIC) codes, uncovering discrepancies and high-risk activities that traditional registries may miss. While GPT-3 alone showed moderate accuracy in business classification, combining it with KPMG's NLP model significantly improved prediction performance, enhancing risk assessment in financial crime investigations. This ensemble approach is intended to support know your customer (KYC) processes, helping financial institutions better identify and mitigate money laundering risks.