Published Paper Details:

This paper presents a new approach to compliance procedures and aims to aid regulators in the process of regulating financial institutions. We propose a model which enables regulators to act as advisors to financial institutions about the expected, acceptable level of behavior. The institution then acts to optimize its business conduct with respect to maximizing profits and meeting the advised level of conduct [1] This is an improvement from the current situation, where advisers to the financial industry provide stepwise regulations (if they do 'A', then they must do 'B' or face penalty) which can be complex, ambiguous, and non-optimal in terms of aligning the interests of the financial industry and the general public. Our model culminates in a system of analytics and a virtual environment which serves as a test bed for market regulations. In recent times, the financial industry has been perceived as an environment dominated by greed, collusion, and corruption [1]. Consequently, financial institutions have been under discussion with regards to compliance and regulations. It is generally assumed that regulators are responsible for the implementation of new regulations and compliance procedures. But in actual fact, compliance procedures and the subsequent penalties for non-compliance are a result of lawsuits filed against financial institutions for alleged fraud cases.Although it is an important development in reg tech, AI learning and monitoring is something that is held as controversial for regulated entities. AI automation has huge potential to improve cost efficiency and at the same time increase the effectiveness of regulation. However, the downside for regulated entities is that it becomes easier to detect wrongful actions and breaches, increasing the probability that they will be caught. This can be seen as an increase in the rate of enforcement and has the potential to create the very regulatory burden that the AI system was trying to alleviate [8]. Decision rules inferred from data can be used to automate and monitor the behavior of the regulated entity. This ranges from passive surveillance, in which the monitoring system alerts a human when the regulated entity has done something that warrants further investigation, to full automation, in which the monitoring system is taking the same actions as it is monitoring on the regulated entity.