Members of the European Parliament’s Committees on Civil Liberties (LIBE) and Internal Market (IMCO) discussed the European Commission’s proposal for harmonised rules in the field of artificial intelligence (AI) (see EUROPE 12898/22) on Monday 21 March, following speeches by several experts. A debate on the same subject was also taking place in Parliament’s Committee on Industry, Research and Energy (ITRE) on the same day.
The discussions in the three committees focused in particular on the issue of definitions.
“There is a growing debate about definitions and scope, sometimes described as too broad, sometimes as too narrow. I think we need to have a horizontal text with as few exclusions as possible. What is prohibited must be clearly defined”, judged the rapporteur of the LIBE Committee, Dragoș Tudorache (Renew Europe, Romania).
For others, like Svenja Hahn (Renew Europe, Germany), it is above all a question of “not overloading the boat”. “We should not add too many layers of legislation, not use definitions that are too broad”, she said.
“If we use definitions that are too broad, we may exclude some areas”, added Deirdre Clune (EPP, Ireland), who, like other MEPs, suggested using the OECD definition of AI.
For the definitions as well as for the issue of the uses of AI systems labelled as ‘high risk’, it will be a question of finding “the balance between innovation and the protection of fundamental rights”, stressed Mr Tudorache.
“Obligations are placed on providers to define high risk AI. This is an interesting method. The Commission’s proposal would cover 10% of current AI systems”, said IMCO Committee rapporteur Brando Benifei (S&D, Italy), while wondering whether a different approach to scope was possible.
A different approach for SMEs
In the ITRE Committee, in addition to the definitions and scope, the need to ‘protect’ small and medium-sized enterprises (SMEs) and start-ups was also discussed.
“Measures to support innovation must be developed. I think we should look at SMEs and start-ups in a more holistic way”, explained Eva Maydell (EPP, Bulgaria), who tabled 126 amendments on, among other things, the development of AI in industry and the responsibility of relevant actors in the value chain.
In addition, several MEPs expressed their concerns about how to make future legislation future-proof against future technological developments and emerging risks.
“We have to be careful that the legislation is not already obsolete by the time it comes into force”, warned Petar Vitanov (S&D, Bulgaria).
In addition, several MEPs also raised the issue - widely debated in Parliament (see EUROPE 12912/35) - of the use of AI systems for surveillance in public spaces.
“People and the protection of fundamental rights must be given priority. Several points are still unclear. We need to focus on applications that can create dangers”, summarised Elena Koundourá (The Left, Greece). (Original version in French by Thomas Mangin)