The EU Artificial Intelligence Act – Bitkom Position for the trilogue negotiations
The Artificial Intelligence Act (“AIA”) of the European Commission (EU COM) aims to build trust in AI, protect fundamental rights and promote innovation and competitiveness. The Council of the European Union (Council) and the European Parliament (EP) adopted their positions on this in December 2022 and June 2023 respectively, and entered the trilogue negotiations in June 2023. As the amendments of the Council and the EP differ strongly from the proposal of the EU Commission in some important points, the upcoming trilogue will be decisive for the concrete design and the balance between risk protection and use of opportunities.
Bitkom therefore again comments on the most important aspects of the AI Regulation and pleads for an appropriate, future-proof and proportionate legal framework for the development, distribution, and use of AI systems. Particularly with regard to the definition of AI, the risk classification, the distribution of roles and responsibilities, the inclusion of so-called General Purpose Artificial Intelligence (GPAI) and the interaction between AIA and existing laws, a reasonable compromise ought to be found.
Our main points on these aspects are the following:
Regulate what makes AI special by concentrating on methods that learn from data
Create a truly risk-based approach by adding a criterion of ‘significant risk of harm’ to the risk-classification
Make sure that interaction with sectorial legislation is frictionless by having consistent definitions and by avoiding duplication of requirements
Focus on cooperation within the value chain to effectively address risks instead of creating requirements for certain AI-systems that do not have an intended purpose
Strengthen the promotion and not only the restriction of AI by creating effective regulatory sandboxes
Make the fast development of harmonised standards a strategic top priority
