The recent Regulation (EU) 2024/1689 (the “AI Act”) lays down obligations for AI system providers in the EU. The AI Act was published on 12 June 2024 and entered into force on 1 August of the same year. The regulation will be fully effective on 2 August 2027, with some provisions becoming applicable from 2 February 2025. It sets for the first time in the world a harmonised legal framework that aims to establish trustworthy AI. The regulation follows a risk-based approach, from production to use, and has the most important impact on AI systems classified as high-risk. Consequently, this new legislation should have a limited impact on the e-commerce sector.
The AI Act subjects providers of AI systems to different obligations depending on the degree of risk that their system(s) pose to individuals (unacceptable, high, limited, or minimal). The classification is established by assessing the threat an AI can pose to fundamental rights, safety, and ethical principles. AI systems presenting an unacceptable risk are fully banned, such as for instance a toy using voice assistance that would encourage dangerous behaviours. When its technology presents a high risk, the provider of the AI system has to comply with a series of strict obligations before and after putting it on the market. For example, an AI is high-risk when used in critical infrastructures, law enforcement, or safety components of products. When the risk is limited, the regulation only defines transparency obligations: AI-generated content must be identifiable, and users should be informed when interacting with an AI.
In the e-commerce sector, as the use of AI is expanding and evolving, going from customer services (like Chatbots) to logistics. It is therefore still difficult to assess the impact of the new Regulation on the future of the sector. A survey published last March by Fevad, regarding the French and European e-commerce sector in 2024, showed that 71% of responding companies have used one or more AI-based solutions, and 91% consider it a highly promising innovation that they plan to integrate into their activity over the next 3 years. Respondents identified marketing (81%) and customer relations (74%) as the areas with the greatest potential in terms of e-commerce. Additionally, certain applications, such as the use of AI for credit-scoring of fraud detection in payments could fall under other category and trigger obligations. In this context, Ecommerce Europe will keep monitoring the repercussions of the AI Act in the sector.
To facilitate the legislative transition for industry, the Commission has established the AI Pact, which entered into force on 1 August 2024. The Pact allows the voluntary stakeholders from the industry to implement the provisions of the AI Act before its full applicability. The Pact follows a two-pillar approach: first, it creates a network for the participants, allowing them to meet and exchange best practices for the implementation of the AI Act; second, it encourages the industry to produce declarations of engagements containing their concrete actions and timeline for the early adoption of the AI Act provisions, including regular reporting on the status of these pledges. The participants are currently in the preparation phase of the pledges, which will be signed on 25 September 2024.
In the meantime, the EU AI Office, which is responsible for developing and coordinating European AI policy, is overseeing its implementation and enforcement. In this context, its board will hold its second high-level meeting on 10 September 2024. Ecommerce Europe will keep monitoring the implementation of this new legislative framework and its impact on the e-commerce sector.