Consumer Protection Safeguards after the AI Act

Consumer Protection Safeguards after the AI Act Assistant professor Dimitrios DEVETZIS1 PhD. candidate Simos SAMARAS2 Abstract The long-awaited introduction of the AI Act, notwithstanding its originality, does not introduce any regulation outside the existing framework. It supplements the safeguards provided in numerous sectors of the EU legislation, inter alia, consumer protection. An examination of the basic EU law along with national legislation of the most influential legal orders, i.e. the French and German one, and model rules of the Draft Common Frame of Reference (DCFR) confirms that the new AI Act sheds light on pre-existing vague legal concepts. In this respect, this novel piece of legislation promotes a better understanding of traditional notions of law applicable to the modern digital reality without introducing totally new rules. Accordingly, certainty of law is reinforced where consumer protection had languished, not because of legislative shortage, but for facts and situations described as necessary conditions of certain rights were contested on the basis of contrary possible interpretations of decisive terms. Ultimately, the very importance of the AI act regarding consumer protection lies not in the introduction of new rights and obligations, but in a multiple practical assistance to the implementation of existing rules adjusting thus the traditional legal concepts with technological development beyond their original scope. Keywords: AI Act, AI Convention, consumer protection, comparative law, contractual fairness, data protection, Draft Common Frame of Reference, EU law, good faith, information accuracy, product safety. JEL Classification: K11, K15, K24 DOI: 10.62768/PLPA/2024/13/2/12 Please cite this article as: Devetzis, Dimitrios & Simos Samaras, „Consumer Protection Safeguards after the AI Act”, Perspectives of Law and Public Administration 13, no. 2 (June 2024): 298-309 Article History Received: 10 April 2024 Revised: 15 May 2024 Accepted: 3 June 2024 1. Introduction 1.1. AI as the subject of regulation An original and innovative initiative can be approached in one of two ways. One possible approach is to prohibit the phenomenon in question from occurring due to a fear of the unknown. Such a treatment is, in the first instance, ineffective, as the threat of sanctions and the potential use of law enforcement in this direction will have little impact if the initiative corresponds to the fulfilment of a certain need and does not conflict with any contrary culture. The alternative approach is one of controlled permission3. Such acceptance, on the one hand, presents few or no obstacles to innovation, while, on the other hand, allows for a significant degree of control in order to prevent abusive or unfair activities. The treatment of the use of artificial intelligence (AI) is similarly subject to this distinction. 1.2. The AI Act and relevant legislative developments 1.2.1. The AI Act After years of deliberation and compromise 4, the European Parliament made the first move 1 Dimitrios Devetzis - Frederick University, Nicosia, Cyprus, . 2 Simos Samaras - National and Capodistrian University of Athens, Greece, . 3 John Frank Weaver, ‘Regulation of artificial intelligence in the United States,’ in Research Handbook on the Law of Artificial Intelligence, eds. Woodrow Barfield and Ugo Pagallo (Cheltenham: Elgar, 2018), 155-56. 4 See documents of the Council of the European Union ST 7536 2024 INIT (Brussels, 18 March 2024), p. 1, 2. Perspectives of Law and Public Administration Volume 13, Issue 2, June 2024 299 by adopting the ‘legislative resolution of 13 March 2024 on the proposal for a regulation of the European Parliament and of the Council on laying down harmonised rules on Artificial Intelligence (Artificial Intelligence Act) and amending certain Union Legislative Acts (COM – 2021 – 0206 – C9-0146/2021 – 2021/0106 – COD)’. The document will become the Artificial Intelligence Act (AI Act) and may be identified until its publication in the Official Journal of the European Union under the number P9_TA (2024)0138. Following this, the process of evolution proceeded rapidly and in a chained manner. Given that the position of the European Parliament was the result of a preceding deliberation with the Council of the European Union and the Committee, the Council approved the resolution without any disagreement. It could be argued that there was still a margin of appreciation by the Council of the European Union and referral back to the European Parliament, according to article 294 (5) TFEU. This was unlikely, as the Council of the European Union would be in contradiction with itself were it to reject the approved text. At this time, the approved text, which was sent on 14 May 2024 as document PE-CONS 24/24, is pending publication in the Official Journal of the European Union. This final version of the AI Act is cited as hereinafter. Nevertheless, the AI Act is no longer the only binding text regarding artificial intelligence. 1.2.2. The AI Convention Despite receiving minimal attention, the Council of Europe drafted a Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law. This was subsequently adopted as text in the 133rd Session of the Committee of Ministers (Strasbourg, 17 May 2024; CM – 2024 – 52 final; hereinafter AI Convention). The future multilateral AI Convention is of greater significance than the AI Act in terms of the countries that are bound by it. Although not yet open for signature, the Convention is addressed to a wider circle of countries than those participating in the Council of Europe. This is explicitly stipulated in Article 30(1). It is notable that powerful economies, including the United States of America, Japan, Australia, and Canada, participated as observers in the preparatory works 5. This capacity allows them to sign the AI Convention. Furthermore, the AI Convention devotes a whole article to the federal states of loose central government (art. 33), which includes the United States of America. These observations lead to the reasonable conclusion that the AI Convention editors directed it to countries far outside the borders of Europe. 1.3. Expectations and results of the regulatory framework It was reasonable to expect that the AI Act would provide for specific rights of consumers and obligations of providers6, given that the economic exploitation of artificial intelligence is at the core of its usefulness. However, in a milieu of opposing views and conflicting interests, it was not anticipated that a clear and unambiguous introduction of new rights to the advancement of consumer protection7 would be forthcoming. While acknowledging the necessity for a later framework, the European legislator is currently focusing on a comprehensive regulatory framework for the circulation of AI products and applications within the EU, as far as the activity of their providers or manufacturers is con (...truncated)