AI-Based RegTech

Businesses in the financial markets are in the midst of a period of overwhelming legislative activity from multiple levels of norm-making in Europe and internationally. The tendency in new legislation has been to introduce ever more detailed control, supervision and enforcement. Indeed, the level of detail in rules and the amount of reporting with which finance businesses must comply has become too vast to survey without automated assistance for the purposes of reporting and compliance management.

This situation has created a surge in the market for advanced data services for compliance management. Applications for compliance with regulation (but also with business codes, contracts, and ethical standards) have become known as RegTech. The most advanced of these applications include artificial intelligence algorithms that make use of machine learning and natural language processing; indeed a 2018 IBM report indicated that these are the most commonly used RegTech algorithms. The most common fields of application for AI-based RegTech are (according to a Deloitte survey) compliance management, risk management, transaction monitoring, mandatory reporting, and identification.

AI-based RegTech is very attractive to businesses struggling to cope with abundant reporting and monitoring. It increases efficiency and reduces costs, serving e.g. bank officers with big data analyses that would otherwise be tremendously time-consuming. For instance, AI-based RegTech helps detect suspicious transactions that could form part of financial crime such as money laundering.

However, AI-based RegTech also raises questions and concerns, perhaps most importantly with regard to the very purpose of reporting standards and internal monitoring. Such measures are primarily intended to raise ethical awareness in organizations and to educate leaders on how to maintain the standards set to protect investors, consumers, and taxpayers. We know that already a significant portion of compliance efforts are handled by algorithms. It may therefore be legitimately asked; are the people in the organizations actually learning anything, or are the algorithms the only ones learning? Some critics even say that the algorithms have been taught how to tick all the boxes and to report impeccable but potentially misleading data, thus covering a reality of decision-making that may (again) grow increasingly reckless. Such concerns serve to illustrate that it is crucial to study the use of AI-based RegTech in the financial sector, to set its uses into the theoretical contexts of risk management and compliance management, and to trace in what ways it actually contributes to (or reduces) transparency and accountability. Thereby we can produce empirical results for an informed discussion on how AI-based RegTech in the financial markets does indeed serve the purposes of legislative and ethical policies, and what uses do not. We can also contribute to designing more efficient means of promoting transparency, accountability, and ethical stringency in the financial markets.

Moreover, we recognize that the phenomena we have identified are not confined to the financial sector, but are present in algorithm-based reporting to e.g. tax authorities and in accounting; uses that are relevant to more complex business structures notwithstanding to which industry they belong. AI has an impact on the organization of any complex business. Consequently this proposal has broad theoretical implications for management and organizational theory, making it a crucial research and teaching topic.