Yesterday, on Friday, EU member states approved globally unprecedented legislation to regulate artificial intelligence after intense negotiations on the balance between maintaining freedom of innovation and security, Agence France-Presse reports.
According to the announcement made by the Belgian President of the Council of the European Union, the ambassadors of twenty-seven countries “unanimously” ratified the agreement reached last December between those countries and members of the European Parliament.
The European Commission presented its project titled “Artificial Intelligence Law” in April 2021, and in late 2022 the emergence of the “GPT Chat” program developed by the California-based startup “Open AI” gave it a new dimension, contributing to the discussions. to accelerate…
This system – similar to other systems capable of generating sounds, images or text – exposes the general population to the vast potential of artificial intelligence, but this technology also comes with various risks, including the publication of false images that look very realistic. is also involved, which raises the possibility of greater possibility of manipulation. In public opinion.
The European Commissioner responsible for this file, Thierry Breton, welcomed the “historic and unprecedented worldwide” legislation, and said: “Artificial Intelligence legislation has generated great interest for the right reasons. Today, countries have recognized the political agreement issued in December. “Approved giving the ideal balance that he has found.” Negotiator between innovation and security.
Paris and Berlin expressed their eagerness to finally legislate to protect emerging companies specializing in artificial intelligence, so as not to prevent the emergence of “European champions” in this field in the future. Diplomats told Agence France-Presse that concerns were taken into account before the text was finalized, so both countries received clarification on its implementation.
German Digital Technology Minister Volker Wissing expressed his happiness “because we have achieved reforms for small and medium-sized companies, avoided inconsistent requirements, and have been able to ensure that we remain internationally competitive.”
For his part, German Economy Minister Robert Habach acknowledged, “This law makes it possible to harness the enormous potential of artificial intelligence while taking into account the risks. In implementing it, we will ensure ease of innovation, legal clarity for companies and Will insist.” The need for non-bureaucratic structures.”
However, the technology world seemed more cautious. “Many of these new rules are still vague and could slow down the development and diffusion of innovative applications,” said Boniface de Chambery, head of European affairs at the CCIA lobby group, which is active in the region. ” and warned. “It will be important to enforce the law well” so that it does not “burden” competitiveness.
Marianne Turdot-Bitker of France Digital, an organization active in the digital sector, said the law “creates significant obligations for startups and small and medium-sized companies, despite some amendments,” and she expressed fear of “additional regulatory barriers.” Benefit American and Chinese competition.
Regarding generic artificial intelligence, rules will be applied to all to ensure the quality of data used in developing algorithms and to verify that they do not infringe copyright law, according to Agence France-Presse.
European regulations require developers to ensure that the sounds, images and texts produced are clearly identified as the product of artificial intelligence, and in particular those related to critical infrastructure, education, human resources and systems at “high risk”. “Increased restrictions on systems will apply. Maintenance, as they will be subject to a series of obligations such as providing control. The implementation of humans, the organization's technical documentation, or risk management systems into machines.
This law provides for specific monitoring of artificial intelligence systems interacting with humans, as well as a duty to inform users about it. As in current European regulations on product safety, the text imposes controls that depend primarily on the companies themselves. The law includes some prohibitions related to applications that violate European values, such as profiling of citizens, mass surveillance systems used in China, or remote biometric identification of people in public places.
On this last point, states have immunity for certain law enforcement functions, such as counter-terrorism. The European Parliament still has to approve the final agreement in the spring, which can no longer be amended, and some rules will come into force six months after its adoption and two years for other provisions.