European agreement for the first legislation on Artificial Intelligence

It couldn’t be the first time, after 24 consecutive hours of confinement, but it was the second time, after 24 hours of rest. The dialogues between the MEPs, the officials of the European Commission and the Council, represented by the Spanish presidency, have reached an agreement this Friday to carry out the first Law on Artificial Intelligence of the planet, which should serve to regulate in the coming years both technologies such as ChatGTP and the use of biometric identification systems in real time and at a distance, such as facial recognition.

“Historic! The EU becomes the first continent to establish clear rules for the use of AI. The law It is much more than a rule book, yesbe a launching pad for EU startups and researchers to lead the global AI race. The best is yet to come,” celebrated the Internal Market Commissioner, Thierry Breton, present in the room these days. “The AI ​​Law is a world first. A single legal framework for AI development that you can trust for the safety and fundamental rights of people and businesses. A commitment that we assumed in our political guidelines and now we fulfill. “I welcome today’s political agreement,” agreed Ursula von der Leyen.

It is not the last step yet, since now both the national governments on their part and the plenary session of the European Chamber will have to ratify the agreement, to see if they accept the modifications that have been made, thus could come into force in a few weeks, if nothing gets stuck. The process began in 2021 with the legislative proposal of the European Commission and slowly advanced the tortuous community path, but when the ball was in the court of the European Parliament, the practical application of the so-called foundational models of generative AI burst in with destabilizing force. the chatbots capable of programming, writing and summarizing books in a matter of seconds and accessible to anyone. This forced a complete rethinking of the philosophy and scope of the directives, and with a sense of absolute urgency.

However, what began with respect, if not fear, and the consensus that there should be very strong regulations has been changing in recent weeks. The European Parliament, which is always (at least on paper) the one that advocates for civil rights, freedoms and is in friction with national capitals, which pay close attention to economic and business sensitivities, came with a list of important requirements. Requiring that a broad list of prohibited uses for AI be included in Article 3, and specific safeguards. The governments, represented by the Spanish presidency, have opted one by one for self-regulation, for prioritize the economic dangers and the fear that Europe will be left behind in this essential technology compared to China or the USand they were inclined to expand the list of risks, which appear in an annex, but avoid prohibitions as much as possible.

“The Council and Parliament reach a provisional agreement on the artificial intelligence law, which provides for ensuring that AI systems marketed and used in the EU are safe and respect fundamental rights and EU values,” said the Spanish presidency, who has been represented throughout the entire process and in these 40 very intense hours by the Secretary of State Carme Artigas.

Failure to comply with the rules, once they come into force in a few weeks or months, can lead to fines ranging from 35 million euros or 7% of global turnover to 7.5 million or 1.5% of business volume, depending on the violation and the size of the company.

Unacceptable risks

MPs argued that AI systems of unacceptable risk are those that are considered a threat to people, such as cognitive manipulation of the behavior of specific vulnerable people or groups: e.g. Voice-activated toys that encourage dangerous behavior in children. Or the social scoring applied in China, social score that serves to classify people based on their behavior, socioeconomic status or personal characteristics. Or biometric identification systems in real time and at a distance, such as facial recognition.

The pressure has been very strong in recent weeks, and especially in the last 72 hours, for MEPs to give in and accept a very lax wording that expands the number of exceptions to the application of the lawgives powers to States to use these instruments in cases of national security and opens the door to biometric categorizations of race, sex or even emotions, in certain circumstances. Dozens of academics have signed a letter since Thursday asking deputies to resist pressures that allow predictive and biometric policing based on race, due to the obvious danger of discrimination. In addition to, for example, exemptions to “a posteriori” remote biometric identification systems, in which identification occurs after a passage of time and which could be used to prosecute serious crimes with judicial authorization.

“Recognizing the potential threat to citizens’ rights and democracy posed by certain applications of AI,” say the MPs in their summary of the agreement, it has been agreed to ban: “biometric categorization systems that use sensitive characteristics (for example, beliefs political, religious, philosophical, sexual orientation, race); the untargeted extraction of facial images from the Internet or CCTV images to create facial recognition databases; the recognition of emotions in the workplace and in educational institutions; the punctuation based on social behavior or personal characteristics; AI systems that manipulate human behavior to circumvent their free will.

“We are the first in the world to establish real regulation for Artificial Intelligence and for the future digital world driven by AI, guiding the development and evolution of this technology in a human-centered direction,” he celebrated with a glass. with champagne in his hand, deputy Drago Tudorache, one of the two rapporteurs parliamentarians.

Marathon sessions

In the first 24-hour marathon, the Spanish presidency ended up putting on the table a proposal that satisfied the Council’s lines, but that some deputies refused to digest. The pause, due to the labor regulations of the institutions, served for the Eurochamber to regroup and ask for a more balanced text.

“For AI systems classified as high risk (due to their significant potential harm to health, safety, fundamental rights, the environment, democracy and the rule of law), clear obligations were agreed. We MEPs have including a mandatory fundamental rights impact assessment, among other requirements, also applicable to the banking and insurance sectors. Artificial intelligence systems used to influence the outcome of elections and voter behavior are also classified as “Citizens will have the right to file complaints about AI systems and receive explanations about decisions based on high-risk AI systems that affect their rights,” the distributed documents say.

Diplomatic sources explained that in the first 24 hours, half of the way had been completed, but there were more than eight important issues left open on the agenda. The big problems were the prohibitions of article 3, biometric identification and exemptions when national security is invoked. The Spanish presidency wants to close the dossier. It is one of his priorities and achieving the agreement can be presented as a great victory, something that marks the balance of the presidency along with the issue of fiscal rules, which is also on the limit. Both elements will be the closure sought by the Ministry of Nadia Calvio in her last days as vice president.

That is why the Eurochamber was not in such a hurry in principle. He reasonably didn’t care about an agreement right now or in three weeks, under the Belgian presidency, much less a shot in the arm. On Wednesday they almost gave up due to exhaustion but stopped. Friday’s day has also been very long, it started at 09:00 and 15 hours later, after several breaks so that the deputies could establish a position among themselves, it was sealed.

Be the first to comment

Leave a Reply

Your email address will not be published.