(Brussels) The new European Union legislation, unprecedented worldwide, to regulate artificial intelligence (AI) officially came into force on Thursday, with the aim of promoting innovation in Europe while limiting possible abuses.
“This is Europe’s pioneering framework for innovative and safe AI. It will foster the development of AI that Europeans can trust. It will help European SMEs and start-ups bring cutting-edge AI solutions to market,” European Commission President Ursula von der Leyen said on social media platform X.
For the most part, this legislation will actually apply from 2026, but some provisions will become binding from next year. It adopts a “risk-based” approach and imposes constraints on different artificial intelligence systems that are proportionate to the dangers they pose to society.
AI systems posing only limited risk will be subject to very light transparency requirements, while high-risk systems, used for example in critical infrastructure, education, human resources or law enforcement, will be subject to enhanced requirements before being authorised in the EU.
These requirements will include, for example, human control over the machine, the establishment of technical documentation, or the establishment of a risk management system. Bans will be rare. They will concern applications that are contrary to European values, such as citizen rating systems or mass surveillance used in China.
The bans “will apply from February 2, with fines of up to 7% of global turnover in the event of violation,” underlines Marcus Evans of the law firm Norton Rose Fulbright, who urges companies to prepare now.
Specific rules will apply to generative AIs like Open AI’s ChatGPT to ensure the quality of the data used in developing algorithms and respect for copyright.
Artificially generated sounds, images and texts will have to be clearly identified as such to avoid manipulation of opinion. The regulation was definitively adopted by the 27 EU member states on 21 May, after its adoption by the European Parliament by a very large majority.
The EU co-legislators reached an agreement on this text at the beginning of December after difficult negotiations, with some countries such as France fearing excessive supervision which would threaten the development of this promising sector.