Giorgia Meloni has announced that artificial intelligence will be at the center of the Italian presidency of the G7 next year. In Europe, however, the ministers of your government express opposing views on one of the key points of the issue on behalf of Italy: How should intelligent algorithms be regulated? And in particular the “foundational AI”, that is, the great linguistic models like the GPT underlying ChatGPT, or those of Google and Meta? According to Business Minister Urso, it is better to let these digital giants regulate themselves. While for the Undersecretary for Innovation Alessio Butti it is not enough: “simple” but “certain” rules are needed.
The document of discord
The negotiations underway in Brussels between the Commission, Parliament and Council on theTO THE Act, the EU legislation which is the first in the world to try to set limits on the development of technology. On the most powerful foundational AI, one of the controversial points, Parliament wants binding rules that guarantee transparency and controls. While a few days ago Italy, France and Germany circulated a document, a “non paper”, which underlines how constraints of this kind would risk penalizing innovation in Europe: they therefore oppose the introduction of “untested rules” and suggest instead of developingmandatory self-regulation through codes of conduct”. Without sanctions for those who go astray.
Butti: “The responsibility for AI is mine”
The document fell like a bomb in the middle of the negotiations, which seemed likely to be closed at the beginning of December, in order to approve the AI Act by the end of the European legislature, but now risks derailing. Urso’s Ministry of Business claims paternity, saying that the common orientation emerged on October 30 during a meeting with his German counterpart Habeck and the Frenchman Le Maire. Too bad that according to the Undersecretary Butti does not represent the government’s position: «Italy has expressed its preference for certain and simplified rules, so as to balance the needs of innovation with ethical principles. However, she did not say she was in favor of codes of conduct because they were not effective enough as they lacked effectiveness. Rather than the “non paper”, reference should be made to the national position expressed so far also during the continuous dialogue with the community institutions”. Behind the difference in views, a conflict of competences emerges, not the first on digital, with Butti claiming the role of “coordination in the field of AI, confirmed on 6 November by the Prime Minister”.
If this is true, the impression in Brussels was different: that of a united front between the three European industrial powers to let Big Tech regulate itself. Just as the sensational and opaque departure from OpenAI by Sam Altmanone of the managers who assured the world of ethics and responsibility in the advancement of technology, seems to support Parliament’s reasons: «It demonstrates the lack of transparency of the companies that develop the most powerful models, which cannot be left to define and execute the rules» , says Brando Benifei, PD MEP and negotiator of the text.
The clash over facial recognition
We will see if the current Spanish presidency will be able to construct a compromise, which is now uphill. In fact, there is also a second front, that of systems that identify people in public places by “observing” them biometric data. Many governments ask for large exceptions for police applications, Parliament is willing to grant them only for serious crimes, subject to authorization from a judge. Here, at least, the Italian position is clear: «In terms of safety, in its various aspects, national prerogatives must be preserved – says Butti – by excluding these areas from the perimeter of the regulation».