A world first. The member states of the European Union approved, on Friday February 2, unprecedented legislation at the international level to regulate artificial intelligence (AI) after intense negotiations on the balance between innovation and security.
The ambassadors of the Twenty-Seven "unanimously confirmed" the political agreement reached in December between the States and the MEPs, announced the Belgian presidency of the Council of the EU.
The European Commission presented its “AI Act” project in April 2021. The appearance at the end of 2022 of ChatGPT, from the Californian start-up OpenAI, capable of writing dissertations, poems or translations in a few seconds, gave it a new dimension and caused the acceleration of discussions.
This system, like those capable of creating sounds or images, revealed to the general public the immense potential of AI. But also certain risks: the dissemination of false photographs, larger than life, has warned of the danger of manipulation of opinion.
If rules targeting AI exist, for example, in China, the European legal framework stands out for its scope. European Commissioner Thierry Breton, in charge of the file, welcomed a “historic, world first” regulation.
“The AI law has unleashed passions, and rightly so! Today, states approved the December political agreement, recognizing the perfect balance found by negotiators between innovation and security,” he said. he declared Friday.
“Stay competitive”
Paris and Berlin have shown themselves to be concerned until the end that the legislation protects start-ups specializing in artificial intelligence, so as not to prevent the emergence of future “European champions”.
Concerns taken into account before finalizing the text, diplomats told AFP. The two countries thus obtained clarifications on its application.
On Tuesday, German Digital Minister Volker Wissing was pleased to have “achieved improvements for small and medium-sized businesses, avoided disproportionate requirements and ensured that we remain internationally competitive”.
Also readCES: AI, the undisputed star of the Las Vegas tech show
The tech world is more circumspect: “Many of these new rules remain vague and could slow down the development and deployment of innovative applications,” lamented Boniface de Champris, Europe manager of the CCIA, a sector lobby, on Friday.
“Good implementation will be crucial” so as not to impose “a burden” on competitiveness, he warns.
The legislation "creates significant obligations, despite some adjustments for start-ups and SMEs", adds Marianne Tordeux Bitker, of the France Digitale professional association, fearing "additional regulatory barriers which will benefit American and Chinese competition".
Copyright protection, human control...
On generative AI, rules will be imposed on everyone to ensure the quality of the data used in the development of the algorithms and to verify that they do not violate copyright legislation. Developers must ensure that the sounds, images and texts produced are clearly identified as artificial.
Reinforced constraints will apply only to the most powerful systems. “High risk” systems – critical infrastructure, education, human resources, maintaining order, etc. – will be subject to a series of obligations such as providing human control over the machine, establishing technical documentation, or the implementation of a risk management system.
Also readChatGPT and copyright: the New York Times, herald of press freedom
The legislation provides for specific supervision of AI systems interacting with humans, with an obligation to inform the user.
As in existing European regulations on product safety, the text imposes controls based primarily on companies.
Exemptions for counter-terrorism missions
Bans will be rare. They will concern applications contrary to European values such as citizen rating or mass surveillance systems used in China, or the 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 fight against terrorism.
The European Parliament must still definitively ratify the final compromise in the spring, which can no longer be modified. Certain rules will apply six months after adoption, two years later for other provisions.
With AFP
The France 24 summary of the week
invites you to look back at the news that marked the week
I subscribe
Take international news everywhere with you! Download the France 24 application