MNB Tests AI in Banks and Insurance Companies

The National Bank of Hungary (MNB) has for the first time carried out a thematic examination of the IT, privacy and other risks of artificial intelligence and machine learning used by banks and insurance companies, the supervision found no violations, but risks that need to be addressed, the MNB told MTI on Wednesday.

According to the statement, the investigation covered three banks and three insurers already using such solutions.

The supervisory authority reviewed the effectiveness of the regulation, monitoring and control systems for the use of machine learning and artificial intelligence (ML-AI). Domestic financial institutions currently use ML-AI solutions mainly in their customer identification, account opening, marketing, customer relationship management (CRM) and campaign management processes.

In the absence of specific legislation, the study was based on the Government Decree on the protection of IT systems of financial institutions, (re)insurers, investment firms and commodity exchange service providers and the MNB Recommendation on IT system protection. Due to the novelty of the subject, MNB has set up a special working group within its organization to develop and apply the new specific examination methodology.

The investigation also covered whether financial institutions had properly handled customers' privacy rights, personal data and special (including biometric) data.

MNB's thematic inquiry did not reveal any risks or shortcomings that would violate current legislation. However, it has made a number of recommendations - sent to institutions in an inspection letter - which, although not in breach of the law, relate to issues that need to be addressed. These were mainly due to the lack or incompleteness of risk analysis, regulation and testing, they said.

The purpose of the thematic study was to enable MNB to understand the current uses of ML-AI, to assess the need for possible supervisory guidance on this issue and to incorporate the methodology into its overall IT supervisory testing methodology. In this way, the examination of ML-AI solutions will be a key focus of further comprehensive MNB investigations.