The National Bank of Georgia approves the regulation on "Data-Driven Statistical, Artificial Intelligence, and Machine Learning Model Risk Management". It is becoming common practice in the financial sector to make business decisions based on the data science models. These models increase efficiency, access to finance, and the quality of decision-making in terms of automation. In response to the current challenges caused by the Covid-19 pandemic, supporting the development of remote and automated banking services is becoming even more important. It is noteworthy that the statistical models and especially the Machine Learning and Artificial Intelligent models may have a low degree of transparency. Consequently, the importance of model risk management increases. It is also necessary to develop the principles of models' ethical application. That is why the establishment of a high standard of model risk management and new approaches for risk mitigation in a financial institution becomes vital.

To encourage the proper use of the models and mitigate potential risks, the National Bank of Georgia has designed the regulation of data-driven statistical, AI, and machine learning model risk management. The purpose of this regulation is to establish a framework to ensure effective risk management. This regulation sets out the basic principles of model development, validation, and application. Model Risk Management standards take into account the existing practices and challenges of the Georgian financial sector and modern international supervisory experience. The document defines the National Bank of Georgia's   authorities over the model risk management process. The requirements of the regulation are built on the proportionality principle and risk-based supervisory approach. Compliance with the requirements set out in the statute will be mandatory for all financial institutions.

The regulation is published on the website of the National Bank at the following address:
https://www.nbg.gov.ge/uploads/legalacts/fin_teqnol/151_04210820.pdf (in Georgian)