In June, the European Insurance and Occupational Pensions Authority (EIOPA) published the report of its Consultative Expert Group on Digital Ethics in insurance providing a set of Artificial Intelligence (AI) governance principles for an ethical and trustworthy AI in the European insurance sector.

The report builds on the EU initiatives over the past years, notably the High-Level Expert Group on AI (AI HLEG) Ethics Guidelines for Trustworthy Artificial Intelligence and the recent European Commission’s (EC) Proposal for harmonised rules on Artificial Intelligence (AI Act).

1. Overview

The report starts by acknowledging the increasing adoption of AI in the insurance sector, as part of a general digitalisation of products and services which was particularly accelerated over the course of the Covid-19 pandemic. Furthermore, the disruptive impact of AI in the insurance sector and its potential in achieving higher levels of accuracy, automation, cost and process efficiency, minimizing risks related to risk assessments as well as creating new products and services is highlighted. To leverage upon these advantages, risks related to the use of the technology should be mitigated, especially those that may hinder: (i) fairness; (ii) non-discrimination; (iii) transparency, and (iv) explainability.
To address these challenges the report identifies the need for robust and future-proof AI governance framework. In this regard, the report provides a set of six (6) governance guidelines taking into consideration the entire AI lifecycle together with non-binding recommendations to insurance entities wishing to design and implement risk-based and proportionality measures by deploying core ethical principles towards a trustworthy AI solution.


2. The Principles

In the table below, you can find a brief analysis of the 6 governance principles outlined in the report:




Main recommendations





  • Carrying proportionate impact assessments
  • Identify suitable and proportionate governance measures on a case-by-case basis
  • Measures promoting ethical and trustworthy AI application

Fairness & non-discrimination



  • Ensure fair and non-discriminatory AI uses
  • Avoid unethical practices (e.g. price and claims optimisation to maximise consumers or their willingness to pay or accept the product)
  • Establish a robust Data Governance
  • Monitoring and mitigating AI biases especially in high-impact AI applications

Transparency & Explainability



  • Understanding the type of business and customer decisions made by AI
  • High AI impact = High Transparency
  • Provide specific to the application explanations

Human Oversight



  • Human oversight throughout the AI lifecycle
  • Carry training on AI

Data Governance & Record Keeping



  • Data Governance should ensure that only appropriate and accurate information is used by AI
  • Monitoring and mitigating biases during collection and processing
  • Compliance with GDPR

Robustness & Performance



  • Robustness of AI-systems and cyber-resilience throughout the AI lifecycle
  • Ongoing performance assessments
  • Development and implementation of AI performance metrics


3. Guidance

As mentioned above, in addition to the principles, the report also offers guidance in carrying an impact assessment of the AI application by analyzing the severity and evaluating the likelihood of the impact to consumers as well as to insurance firms (Annex 2). In this regard, the report provides an analysis of the benefits and ethical challenges of various, commonly used AI applications in the sector, from market research to provision of e-services and claims management (Annex 1). It should be noted that the impact assessment framework proposed, is also taking into consideration the data protection impact of the application to the data subjects, pursuant to GDPR.


4. Conclusion & Next Steps

Considering that at this moment there is not a concrete regulatory framework for AI, neither horizontal or sector-specific, the EIOPA report should be welcomed. Although it is not binding, since it is not a formal recommendation, it is expected to pave the way of future initiatives of the Authority in the area.
At the same time, considering that the governance principles identified in the report have been repeatedly recognised in reports, recommendations and initiatives internationally and in the EU, it is expected that they will be formalised in future policy initiatives. Therefore, the report provides an opportunity for insurance entities operating in Europe to jump start in designing and implementing their AI governance framework and assessing the conformity of their AI solutions in the light of the governance principles.
At VdA, our team of experts is delivering strategic and sophisticated legal advice to help our clients in the insurance sector to prepare for the challenges on the new EU regulatory framework for AI-systems and to help them built a robust and futureproof AI Governance.