AI Act: PFUE seeks more flexibility for regulatory sandboxes

The French Presidency of the EU (PFUE) has proposed new changes in the part of the law on artificial intelligence (AI) relating to innovation and regulatory sandboxes (“regulatory sandboxes”) in a final compromise text before the meeting of the telecommunications working group on Friday (June 17), the body of the Council responsible for taking work forward on a technical level.

While leading the Council of the EU, France made progress on several parts of the file through partial compromises, and it is now handing over to the Czech Republic, which wishes to achieve a general approach under its leadership. .

In this new text, the French government has returned to parts of the proposal that are particularly close to its heart, namely the provisions relating to regulatory sandboxes, which are controlled environments set up by a regulator in which private actors can experiment and test new products under special and limited exemptions.

As the text was sent relatively late and Friday’s meeting being the last meeting of the PFUE working group, the compromise will probably only be briefly discussed and no substantial changes are expected, according to an EU diplomat.

The article about AI regulatory sandboxes has been “significantly shortened and simplified in order to provide more flexibility to Member States when establishing these sandboxes”reads the introductory memo to the compromise document obtained by EURACTIV.

The text returns to provisions previously proposed by the PFUE to avoid imposing too many constraints on national authorities by removing or watering down their obligations, for example on the aspect of reporting or cooperation with other authorities.

A revealing element of this general tendency is that the reference to “common general rules” for regulatory sandboxes that the European Commission would establish under secondary law has been removed in favor of more vague wording on “great common principles”.

This more flexible approach is in contradiction with that proposed to the European Parliament by the Committee on Industry (ITRE). Paradoxically, Eva Maydell, ITRE’s draftsman, belongs to the European People’s Party (EPP), the conservative group which has so far sought to align itself more with the Council.

Spearheaded by Ms. Maydell, the ITRE report includes a brand new appendix that provides guidelines on what sandboxes should offer AI vendors to streamline them across the block by giving the European Commission a coordinating role.

The French government has also added a new paragraph to give AI vendors additional benefits for their participation in regulatory sandboxes, namely that the tests performed in the sandboxes will be taken into account by the authority or the competent body which will assess the conformity of the system with the regulation.

At the same time, AI providers will not be able to request a regulatory sandbox, and a new provision allowing real-world testing under regulatory sandboxes has been included, with several Member States considering that was not clear enough in the original proposal.

Previous changes to the legal basis for collecting personal data have been removed “due to the lack of legal certainty presented by this provision which could be challenged in court as not being in conformity with the GDPR. »

The possibility of setting up a regulatory sandbox aimed at increasing the proper understanding of a certain technology by the authority has been removed from the text, as well as the mention that participation should be free and limited to a maximum of two years.

The list of public interests for which personal data may be subject to additional processing in the context of a ” Sandbox “ has been extended, adding climate change, sustainable energy and cybersecurity.

Regarding testing of high-risk systems under real-world conditions, the relevant articles have been amended to clarify that such testing takes place outside of regulatory sandboxes.

The new compromise text has also returned to the part relating to fines, with the aim of clarifying the number of penalties due to the different types of infringement. Fines for EU institutions and bodies have been aligned with those for other organisations.

We wish to say thanks to the author of this article for this remarkable content

AI Act: PFUE seeks more flexibility for regulatory sandboxes

Explore our social media accounts along with other related pages