ARTICLE1 March 2023

Call for the European Parliament to govern the use of AI based on risk assessment

Proposed AI Act: the rapporteurs and shadow rapporteurs appointed by the European Parliament are currently working hard to finalise a compromise text; one which will greatly influence the final legislative text. Several amendments to the draft AI Act are presently under discussion. These would result in far-reaching over-regulation of artificial intelligence, weakening possible economic and societal development as a result. One of the key elements of the original proposal is a clear, risk-based approach for various AI systems and uses.

AI will be a key technology for economic development and the AI Act will shape how it will be developed and used in Europe. The Confederation of Swedish Enterprise, along with its sister organisations within BusinessEurope, has supported the European Commission’s risk-based approach. It is crucial that the AI Act is based on a clear risk-based assessment of different systems and use cases. This will ensure that the AI Act is relevant and that the use of AI in the EU is not over-regulated.

The European Parliament is now discussing a number of amendments that differ significantly from those originally proposed in the AI Act. It is essential that any new requirements and amendments that are made take into account their technical feasibility, their impact on legal certainty and the ability of AI developers, distributors and users to comply with them. The rapporteurs and shadow rapporteurs appointed by the European Parliament are currently working hard to finalise a compromise text; one which will greatly influence the final legislative text.

Therefore, in a statement to the European Parliament, the business community is calling for allocating the time required the necessary time to finalise a compromise text that supports innovation and the use of AI in Europe.

The key areas for focus are the need to:

Maintain the risk-based approach set out the original proposal while defining ‘high risk’ in a way that reflects this.
Allocate responsibility along the value chain and provide the contractual freedom to do so.
Include, in Annex III, only those uses that are truly high risk.
Rely on a definition of AI that is recognised by industry, is in line with the OECD definition and does not include anything other than AI itself.
Provide legal certainty for the extraterritorial application of the scope of the law and its impact on international trading partners.
Ensure that the AI Act does not duplicate or create competing obligations with existing legislation.
Avoid uncertainty for investment, trade and innovation.