The European Parliament gives its final green light to the AI Act, the world's first law on artificial intelligence. The measure was approved today in plenary session with a very large majority: 523 votes for, 46 against, 49 abstentions. “Democracy 1 – Lobby 0”, rejoices the European Commissioner for the Internal Market, Thierry Breton, on social networks. With today's vote, the European Parliament managed to approve one of the most anticipated regulations of this legislature before the June elections, which effectively made the European Union the world's leading economic power in launch an ad hoc system of rules for artificial intelligence. The aim of the new provision, we read on the European Parliament website, is to “protect fundamental rights, democracy, the rule of law and environmental sustainability against high-risk AI systems”. Now that the law has been definitively approved by Parliament, all that remains is approval by the Council and final verification by lawyer-linguists. Barring any surprises, the text will therefore be adopted before the end of the legislature.
What does the regulation provide?
The path traced by Brussels with the new AI law follows a very clear approach: regulate the artificial intelligence sector without stifling its development. The new rules therefore aim to ban certain AI applications that threaten citizens' rights. Some examples: biometric categorization systems based on sensitive characteristics; blind extrapolation of facial images from the Internet or CCTV cameras; emotion recognition systems in the workplace and in schools; social credit systems; predictive policing practices, which involve the use of artificial intelligence systems by law enforcement to predict the location, method and perpetrator of a future crime. With the AI law, police forces will only be able to use biometric identification systems under certain conditions.
Obligation of transparency
The other big novelty of the AI Act concerns transparency. AI systems will in fact have to meet a series of minimum requirements established by law, while citizens will be recognized the right to file complaints or receive explanations on decisions based on artificial intelligence systems . All AI systems will also have to respect European copyright rules in the training phases of the different models. It is also a highly debated subject in the United States, where recently the New York Times he sued OpenAI and Microsoft, accusing both companies of using thousands of his articles in blatant copyright infringement.
The search of Benifei: “Italian government absent”
The rapporteur for the measure approved today in Strasbourg is the Italian Brando Benifei, group leader of the Democratic Party in the European Parliament. “After two years of intense work, we have finally managed to approve the world's first binding law on artificial intelligence,” rejoiced the Democratic MEP after the approval of the law. However, there is also criticism of the executive led by Giorgia Meloni: “I believe that today, in addition to providing answers on the question of resources, the Italian government should work on implementation of the regulations, because in my opinion – Benifei clarified – during Unfortunately, the negotiations have been very absent and sometimes even confusing. We have seen arguments between ministers which have not benefited our government's negotiating work here. »
Cover photo: EPA/Ronald Wittek | A debate during the plenary session of the European Parliament (Strasbourg, March 13, 2024)
Read also:
