Our researcher Susanna Lindroos-Hovila tells about the EU's proposal for regulating artificial intelligence in an article by the University of Helsinki's science corner on 14.9.2021.

Below is a quote from the article published in the science corner.

1. Regulation targets artificial intelligence products

The Commission sees regulation primarily as product regulation. The legislation is intended to restrict those systems that can be used on the EU's internal market and brought here. The aim is to increase citizens' confidence in products, i.e. artificial intelligence systems within the EU and to create as functional and free internal markets as possible. 

Unlike the much-discussed data protection regulation GDPR (General Data Protection Regulation), the regulation of artificial intelligence does not start from human rights or their protection – however, it is one part of the package.

2. Regulation is based on risks

 In the legislation, artificial intelligence systems would be divided into prohibited, high-risk and low-risk systems. The classification would be based on what kind of harm these systems cause to individuals and society.

Prohibited systems according to the proposal include, for example, systems that alter human behavior in a harmful way, exploit the weakness of a particular group, or that score people.

High-risk systems include, for example, traffic infrastructure, safety-related components (including robotic surgery) and public administration systems that are of great importance to people.

Most of the regulation would concern high-risk systems. For these, the Commission proposes certain requirements that systems must meet. Otherwise, they could not be developed for the EU market.

3. One artificial intelligence regulation for the entire European Union

The new rules would be binding as such throughout the European Union. This is because the Commission proposes legislation specifically in the form of a regulation. 

EU member states would have little legislative authority of their own, meaning that practically the entire artificial intelligence field would henceforth be subject to EU legislation. 

The aim is for the same rules to apply to artificial intelligence in all member states, which would also make product supervision easier. The Commission thus sees a future with only one common artificial intelligence market throughout the European Union. 

What happens next?

The Commission's proposal is just a proposal. It will be considered in the normal EU legislative process. After that, all member states will negotiate on the content of the rules. The EU Parliament must also approve the regulation before it can come into force.

https://www.helsinki.fi/fi/uutiset/tekoaly/tarvitseeko-tekoaly-kansainvalisia-saantoja-nama-kolme-asiaa-eun-ehdotuksesta-tekoalyn-saantelemiseksi-tulisi-tietaa