Table of Contents: I. Introduction. – II. The evolution of the “New Approach” to harmonisation in the EU: setting the scene. – II.1. The “New Approach” to technical harmonisation. – II.2. The Digital Single Market. – III. Harmonisation in the Draft AI Act: old wine in a new bottle, or new wine? – III.1. Risk-based approach of the Draft AI Act. – III.2. The “New Approach” technique in the Draft AI Act. – III.3. “Privatization” of AI Regulation by ESOs and conformity assessment bodies. – IV. Future-proofing AI regulation through the “New Approach”. – IV.1. The future-proofness from the perspective of the Internal Market. – IV.2. Future-proofness from the perspective of fundamental rights and ethics. – IV.3. The future-proofness from a democratic and legitimacy perspective. – V. Conclusion and recommendations.
Abstract: In April 2021, the European Commission proposed a Regulation on Artificial Intelligence (AI) as part of a package of EU legislative harmonization measures that seek to tackle the societal challenges of digitalization and technological innovation. The proposed legislation draws heavily on the “New Approach” technique for the technical harmonization and standardization of goods. This raises several questions and concerns. Firstly, it can be questioned whether a harmonization technique that has been developed for health and safety standards in the offline, physical market, can be that easily transposed to the field of AI, where ethical and fundamental rights issues abound. Secondly, despite its success, the “New Approach” regulatory technique has been subject to much criticism, such as the responsibility of the manufacturers to carry out a conformity assessment, the role and decision-making powers of the private law standardization organizations and notified bodies, and the lack of public participation and public oversight in standardization and certification processes. These concerns are aggravated in the AI environment due to the pertaining legal, ethical and fundamental rights issues. This Article therefore seeks to explore the future-proofness of the proposed AI Act from three perspectives: i) the internal market; ii) protection of fundamental rights and iii) democratic legitimacy in the EU decision-making processes. Ultimately, it offers a broader reflection on the policy and legal implications of AI and proposes a number of recommendations on how to increase the fitness of the future AI Act, bearing in mind the balance between the economic and fundamental rights goals of the AI Regulation.
Keywords: EU Digital Single Market – AI Act – standardisation – “New Approach” – harmonised standards – fundamental rights.
European Papers, Vol. 8, 2023, No 2, pp. 583-610
ISSN 2499-8249 - doi: 10.15166/2499-8249/677
* Professor of Public Economic Law, Utrecht University, email@example.com.
** Assistant Professor of Public European Economic Law and Technology, Utrecht University, firstname.lastname@example.org.
*** LLM Graduate in European Union Law, Utrecht University, email@example.com.