Ethical artificial intelligence governance principles - EIOPA report

On 17 June 2021 the report “Artificial intelligence governance principles: Towards ethical and trustworthy artificial intelligence in the European insurance sector” (the “Report”), drafted by the EIOPA's Consultative Expert Group, was published. The Report provides six principles that insurance sector entities (both insurance undertakings and insurance intermediaries) should keep in mind when implementing AI-based solutions in their organisation, as well as when continuing to use these solutions. The principles indicated in the Report are not binding. However, they can certainly be treated as an indication of the direction of changes that will affect the insurance market in the future. The EIOPA-organised event, in connection with the Report, was also held on 20-24 September 2021. The discussions held during the meetings at this event confirm that the use of artificial intelligence by entities in the insurance sector will involve many challenges.

Principle of proportionality

The use of artificial intelligence should be adapted to the nature, scale, and complexity of the activities of insurance undertakings and insurance intermediaries. The governance measures applied should be proportionate to the potential impact of artificial intelligence on these entities and their clients.

Principle of fairness and non-discrimination

This principle is primarily related to preventing the restriction of access to insurance for so-called "sensitive" clients (e.g. people with physical or mental disabilities), transparency in data processing, or taking steps to monitor, reduce, and remove biases in training and testing data, as well as in the AI-based systems themselves.

Principle of transparency and explainability

Insurance undertakings and insurance intermediaries should take measures to ensure that the operation of AI-based systems, as well as their performance results, are transparent and explainable to the greatest extent possible. People who interact with an AI-based system should be aware of this and be properly informed. Information should be made available to them about the way the system has been designed, the variables and parameters used by it, as well as its limitations.

Principle of human oversight

A human should supervise the operation of an AI-based system at every phase of the operation of that system. In this regard, insurance undertakings and intermediaries should ensure that appropriate procedures and the division of roles and tasks in the organisation are in place.

Principle of data governance of record keeping

The data used by the AI-based system should be accurate, complete, and adequate. Data along with modelling methodologies should be stored by insurance undertakings and intermediaries in a secure environment that guarantees access to data only to authorised individuals. Data management should take into account the requirements provided by legal acts on personal data (e.g. GDPR).

Principle of robustness and performance

A robust and efficient system is one that works reliably and does not cause damage - both in the technical and ethical aspects. To this end, insurance undertakings and intermediaries should continuously monitor the performance of systems and take appropriate measures. Furthermore, it is essential that these entities employ resilient IT systems and infrastructure.

We encourage you to contact us for more information about the Report's impact on insurance undertakings and insurance intermediaries.