It is the first binding law in the world on artificial intelligence (AI). The European Parliament adopted rules on Wednesday March 13 to regulate the use of this technology. The text, which aims to protect rights and security, while supporting innovation, was voted by a very large majority by MEPs meeting in plenary session, with 523 votes in favor and 46 votes against.

The European Commissioner in charge of the file, Thierry Breton, welcomed on X the “massive support” of Parliament for this text which establishes the “first binding and comprehensive rules in the world for reliable AI”. The bill was presented by the European Commission in April 2021.

At the end of 2022, the appearance of ChatGPT, the Californian OpenAI start-up capable of writing essays, poems or translations in a few seconds, gave it a new dimension. This system revealed the enormous potential of AI, but also its risks. The distribution of false photos or videos, larger than life, thus alerted to the danger of manipulation of opinion.

A “historic” day

“Today is a historic day on our long road to regulating AI,” said the text’s co-rapporteur Brando Benifei (Social Democrats) to the press on Wednesday morning. With this text, “we have managed to find a very fine balance between the interest in innovating and the interest in protecting,” added co-rapporteur Dragos Tudorache (Renew, centrists and liberals).

However, this legislation “is only the beginning,” he noted, recalling that artificial intelligence continues to evolve rapidly. “We will have to be very attentive to this evolution of technology in the future and be ready to respond to the new challenges that may arise,” warned Mr. Tudorache.

The legislation provides for a two-tiered approach. “General purpose” AI models will have to comply with transparency obligations as well as European copyright rules. As for systems considered “high risk”, used, for example, in critical infrastructure, such as education, human resources or law enforcement, they will be subject to stricter requirements.

The text also prohibits citizen rating or mass surveillance systems used in China, or even remote biometric identification of people in public places. On this last point, States have however obtained exemptions for certain law enforcement missions, such as the prevention of a terrorist threat or the targeted search for victims.

“Fuzzy rules”

European legislation will be equipped with means of surveillance and sanctions with the creation of a European AI office, within the European Commission. It can impose fines ranging from 7.5 million to 35 million euros, depending on the offense and the size of the company.

“We regulate as little as possible, but as much as necessary,” European Commissioner Thierry Breton wrote on X. The Twenty-seven should approve the text in April before the law is published in the Official Journal of the European Union in May or June.

But the tech world has been cautious. “Many of these new rules remain unclear and could slow down the development and deployment of innovative applications,” estimates Boniface de Champris, Europe manager of the CCIA, a sector lobby.

For their part, the Multinational Observatory (France), Corporate Europe Observatory (Belgium) and LobbyControl (Germany) fear that lobbies will weaken the implementation of the rules surrounding AI. “Many details of the AI ​​Act remain open and need to be clarified (…), for example regarding standards, thresholds or transparency obligations. The composition of the advisory board of the new European AI agency also remains unclear,” they warned in a joint statement.