Banks: Considering AI and machine learning in model risk management

Banks and financial service providers have been using models for decades, for example to manage risks or calculate prices. There has always been a risk that incorrect models or the incorrect use of models can lead to decisions being made that can have negative consequences, such as financial losses. This risk is also referred to as model risk.

In order to measure and reduce model risk, banks have established extensive and complex approaches to model risk management (MRM). However, with the increasing use of artificial intelligence (AI) and machine learning (ML), a comprehensive adaptation of this model risk approach is necessary.

In addition to the undisputed advantages, AI/ML approaches bring with them new types of risks of their own that are not yet sufficiently taken into account in existing model risk management frameworks. It is also to be expected that AI/ML models will increasingly be used in areas that were previously not supported by models, or only to a limited extent. In addition, there are growing regulatory requirements such as the EU Artificial Intelligence Act.

In order to meet the new challenges posed by AI/ML models and their regulation, it is advisable to adapt or expand specific aspects of the existing MRM approach. These adaptations make it possible to use existing methods, processes and tools and thus generate synergies. In the white paper "Modern Risk Management for AI Models", we derive seven fields of action for adapting the model risk management framework based on the specific requirements of AI/ML algorithms.