Artificial Intelligence: the Commission’s Regulation Proposal and AI strategy in Luxembourg

Introduction

Artificial Intelligence (“AI”) is developing rapidly. AI-based systems and technologies are expected to bring far-reaching social and economic benefits for individuals as well as for companies and other players in many industries such as healthcare, farming, education, energy, logistics, justice, climate change mitigation etc.¹  AI may, however, also entail a number of potential risks and generate harm to the health, safety or the fundamental rights of citizens. In light of these opportunities and associated threats, the European Commission has identified the need for a regulatory framework on AI, which is the very first of its kind.

What’s new?

On 21 April 2021, the European Commission published a Proposal for a Regulation on Artificial Intelligence² (the “AI Regulation Proposal”) which constitutes a decisive step in its continuous strategy for artificial intelligence aimed at putting European values at its centre.³  In this context, as a follow-up to the Guidelines for Trustworthy AI and the White Paper on AI, the AI Regulation Proposal aims “to turn Europe into a global hub for a trustworthy AI”.⁴  At the same time, Luxembourg pursues its national strategy on AI by clarifying its position towards AI systems and subsequent (economic and business) opportunities.

Definition of AI systems

The AI Regulation Proposal provides first for a future-proof⁵  definition of AI systems qualifying them as: “software that is developed with one or more of the techniques and approaches listed in Annex I and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with.”⁶

Territorial and material scope

The scope of the AI Regulation Proposal is broad as it is planned to apply to:

1. 1) Providers (whether they are public or private players) who offer and who place AI systems on the European market, irrespective of whether they are established within or outside of the European Union.
2. 2) Users of AI systems located within the Union.
3. 3) Providers and users of AI systems established outside the EU where the output produced by the AI system is used in the EU.

Categorisation of AI systems

The AI Regulation Proposal classifies AI systems in categories depending on the risk of potential harm they can have towards individuals and their fundamental rights. In light of this, AI systems carrying a so-called unacceptable risk will be prohibited, as they may constitute a clear threat to the safety, livelihoods and rights of people. AI systems carrying a high risk will be subject to a stricter regulatory regime with respect to security and accountability obligations. Finally, AI systems bearing a limited or minimal risk will have to respond to certain transparency criteria, which are all outlined in the AI Regulation Proposal.

Sanctions

In the event of non-compliance, the AI Regulation Proposal also foresees administrative fines, which can reach up to EUR 30,000,000 or 6% of the total annual turnover, whichever is higher.

Creation of a European advisory body and national supervisory authorities

A European Artificial Intelligence Board at the level of the European Union, which will be composed of representatives from all EU Member States (including Luxembourg) and the European Commission, would be established, to advise and assist the Commission in connection with the AI Regulation. At national level, Member States must designate national competent authorities and a national supervisory authority responsible for the enforcement of the regulation and to provide further guidance and advice.

Luxembourg’s position and its AI strategy.

In 2019, Luxembourg launched its strategy on AI with the ambition to become one of the most advanced digital societies in the world, especially in the EU while supporting human-centric AI development.⁷  In that context, on April 28, 2021, Prime Minister Xavier Bettel, also Minister of Communications and Media presented the results of a public consultation on AI.⁸  For that purpose, 20,000 Luxembourg residents of the age of 16 years and over were consulted via a survey developed by Luxembourg Institute of Socio-Economic Research (LISER).

The results of the consultation showed that despite the fact that a large majority of Luxembourg citizens consider AI as a facilitator of daily life tasks (70%) and repetitive work-related tasks (64%), a certain scepticism towards AI nevertheless persists. The mistrust resides in the fact that AI may be unable to distinguish between good and bad consequences, or may eventually act in a discriminatory or biased way. However, it appears that 77% of the consulted population have solid confidence in the use of AI in the public administration, especially to reduce paperwork, but also to monitor mobility or to receive accurate medical diagnoses in the context of disease prevention.

While this public consultation gives a fair overview of how AI is currently perceived by the population, the Prime Minister emphasised the importance of an absolute respect of human rights during its development and use.

The impact of the AI Regulation Proposal

These rules will provide Europe with a leading role in setting a global standard with regard to the regulation of AI systems. In the meantime, the AI Regulation Proposal will be subject to the ordinary legislative procedure of the European Parliament. After being adopted, the AI Regulation Proposal will become a directly applicable regulation across the Member States of the European Union. Companies located outside of the European Economic Area will also have to comply with these new standards if their AI systems output affect European citizens. We will therefore remain alert on questions regarding the impact on different economic players worldwide.

As to Luxembourg, we will monitor how the government will position itself in its continuous strategy on AI, during the implementation process of the AI Regulation Proposal.

For any questions please contact the ICT, IP, media and data protection team:

Linda Funck, Partner | Tel: +352 44 66 44 5164 | E-mail: lindafunck@elvingerhoss.lu

This may also interest you:

• Opinion of the EDPS on the Commission’s White Paper on Artificial Intelligence