Until the end, the negotiation was fierce. Started on Wednesday December 6, the final discussions between negotiators from the European Parliament and the European Council to regulate the use of artificial intelligence (AI) on the Old Continent only ended on Friday shortly before midnight, after more than thirty- five hours of discussions.
This new regulation, called Artificial Intelligence (AI) Act, could be a reference in the world, while generative artificial intelligence services are emerging at high speed capable of developing, according to requests made by users, texts in natural language, but also images, videos, songs… The very ones which did not exist at the start of the discussions, proof of the speed of technological changes.
On these generative AIs, the compromise provides for a two-speed approach. Rules will apply to everyone to ensure the quality of the data used to feed the algorithms and proper respect for copyright. Developers will also have to guarantee that the sounds, images and texts produced will be clearly identified as artificial.
Rare bans
Reinforced constraints will apply to systems judged to be “high risk”, mainly those used in sensitive areas such as critical infrastructure, education, human resources, law enforcement. A series of obligations will be imposed on them, such as providing for human control of the machine, establishing technical documentation, or even setting up a risk management system.
Bans will be rare. They will concern applications contrary to European values such as citizen rating systems, behavior manipulation or mass surveillance, or the remote biometric identification of people in public places, except for exceptions linked to security, such as for fight against terrorism.
This text, which will come into force no earlier than 2025, will be accompanied by the creation of a European AI office, and will make it possible to financially sanction companies that contravene it, up to 7% of turnover, capped at 35 million euros. European citizens will have the capacity to lodge complaints themselves.
At the end of the negotiations, Thierry Breton welcomed an agreement ” balance “. ” It’s too complicated ; it’s going to take too long; it goes against innovation, let developers self-regulate” : these are the criticisms that the European Commissioner for the Internal Market says he has suffered, convinced on the contrary that the agreement concluded will serve as ” launching ramp “ for European companies invested in artificial intelligence, within a stable legal framework.
You have 45% of this article left to read. The rest is reserved for subscribers.