Can we keep up with the machines? Stronger and faster artificial intelligence systems require robust risk management practices

No longer just an issue of isolated enterprise, regulatory or reputational risk for financial institutions, compliance failures are indicators of potential systemic deficiencies that can frustrate the mission and ethical goals of a firm. What is more, compliance failures may impede and compromise a financial institution's ability to deliver core financial products and investments. Recent advancements in data management and computing capacity have ushered in a wave of business technology solutions that rely on the power of artificial intelligence (AI) to transform vast quantities of data into useful business and risk management information. Financial institutions utilise these technologies to predict behaviour, make decisions, identify threats and meet regulatory requirements. An unintended consequence of the proliferation of Big Data and advanced analytics is the concomitant expansion of AI-driven models that tend to amplify social and economic biases. As AI-based technologies expand across compliance and risk management functions, they must be subject to rigorous examination and testing. Robust model governance must be a core component of every financial institution's overall risk management and corporate governance strategies. The extent of a financial institution's model governance must align with the extent and sophistication of its model use. This paper sets out the regulatory trends related to AI in compliance and risk management applications and the risks associated with inadequate data management, over-automation and other risk management oversight failures. The possible adverse outcomes are illustrated by means of a case study relating to the detection of money laundering associated with human trafficking. Recommendations for model risk management and model governance follow.