As EU policymakers have made their way back to Brussels, it is time for us to give an update on the latest developments on the Artificial Intelligence (AI) Act. Since our latest update on the file, both the Council, now chaired by the Czech Presidency, as well as the European Parliament (EP) have made substantial progress on the file.
To provide a brief summary of the AI Act, the European Commission with this legislative proposal seeks to regulate the rapidly developing AI technology through a risk-based approach, aiming to boost the incitement for innovation, while simultaneously ensuring trustworthy use of AI, safety and fundamental rights as the heart of the matter. With the proposed regulation for AI, the EU aims to become a global leader of future-proof AI technologies. Within the e-commerce sector, online retailers are mainly users of AI systems that they exploit to provide an experience more tailored to customers’ needs and improve internal efficiency. At the same time, online merchants increasingly invest in the development of AI solutions. In both cases, the use of AI in e-commerce generally offers benefits to consumers, businesses, and society, while carrying very little to no risk. Ecommerce Europe therefore carefully follows the development and debate on the field of AI legislation and engages with policymakers to ensure the opportunities for AI use and innovation in retail remain protected.
The Czech Presidency carrying on the baton
Over the summer, the Czech Presidency has presented the second compromise text of the Council, in which it attempts to settle previously contentious issues, namely how to define an AI system, classification of AI systems as high-risk, governance and enforcement, as well as national security exclusion. Among the changes suggested by the Czech Presidency is a narrower definition of AI systems, providing for a clearer distinction between AI systems and more traditional software systems. Moreover, the Presidency’s compromise text introduces a new horizontal framework to complement the classification of high-risk AI in order to ensure that innocuous AI systems will not become subject to the new regulation once adopted. Regarding the aspect of governance and enforcement, the Czech Presidency proposes strengthening the mandate of the European Artificial Intelligence Board and further calls on the Commission to ensure facilitation of independent technical and scientific counselling for said Board.
Since the publication of the Czech Presidency’s second compromise text, Member States have had until early September to provide feedback. This input will be used to prepare the next compromise text, which is to be expected later in September.
Smooth progress in the European Parliament, but deferral of controversies
As addressed in our previous update on the AI Act, the EP Committees IMCO and LIBE, jointly responsible for this policy file, received more than 3000 amendments for the AI Act draft report. Based on these, the two co-rapporteurs spearheading the negotiations on behalf of the EP, Dragoș Tudorache (RE, Romania) and Brando Benifei (S&D, Italy), shared their fourth batch of compromise amendments on 1 September. This batch of compromise amendments were further discussed during a technical meeting on 6 September, where discussions progressed smoothly, as no particularly sensitive issues were put to the table. Seeking to find compromise on the least controversial aspects of the proposal first, the more contentious amendments, e.g., regarding regulatory sandboxes and their interaction with general-purpose AI systems, have thus been postponed for a later stage. The Committee vote of the joint IMCO-LIBE report will take place at the end of October 2022.