Adnkronos Milan, 17 June (Adnkronos) – Artificial intelligence remains one of the topics under the spotlight of the EU. With the Ai Act in the implementation phase, the debate remains very heated: doubts concern potential changes or even the postponement of the implementation of some key points, with crucial elements such as the guidelines and the "code of practice" still missing.
The first phase saw a real implementation on February 2, 2025, prohibiting those artificial intelligence systems cataloged by "unacceptable risks" and introducing requirements for companies operating on the European market. The next steps are still being defined, and this was precisely what was discussed during the round table "Ai governance between innovation and rules: a European challenge" organized by Meta in collaboration with Ispi in Milan. In comparison, speakers from the institutional and regulatory world, and innovation and businesses.
Andrea Bertolini, director of the Center for the Regulation of Robotics and AI (Scuola Superiore Sant'Anna), opened the debate with a very critical intervention regarding the AI Act, as currently formulated. “Artificial intelligence is an extremely ambiguous term,” he said, “and the classification by risk levels cannot overcome this ambiguity. The damage resulting from conceptual ambiguity can be seen in the definitions of prohibited applications and high-risk systems.” “The guidelines do not give elasticity to the system and, since they are not binding, they create uncertainty. Not having absolute clarity on the technical issues means having divergence in the application.” Edoardo Raffiotta, professor of artificial intelligence law at the University of Milan Bicocca, has a different opinion: “Calling the AI Act responsible for the difficulty of innovation in Europe is unfair.” The professor stressed that the real problem is excessive regulation: “Let's clarify which regulations need to be reviewed or even repealed. In the meantime, we can see the many state-level regulations that are emerging in the United States. In the US, there will be 48 regulations out of 50 states, there too there is no regulatory uniformity but this does not impede innovation and growth”.
The legislation establishes a comprehensive regulatory framework that applies to any organization involved in the development or implementation of AI systems that can affect users or markets within the European Union. The regulation aims to address any distortions in the use of AI. The main objectives include making AI safer for public and commercial use, mitigating security risks, ensuring human control over AI systems, protecting data privacy and ensuring transparency in applications that use this technology. The set of rules has enormous geopolitical potential and was created with the aim of making Europe the center of digital soft power, given the role it would carve out for itself as the first body to implement systemic regulation on AI systems. This is far from a neutral effort, in a historical period marked by transatlantic tensions. Globally, two approaches coexist: on the one hand, the AI Act, a risk-based model, while in the United States, liability is ex post, with market-driven rules and a principle of least intervention whereby innovation comes before regulation. The consequence is the emergence of divergent standards and regulatory fragmentation that further complicates transnational operations.
For Luca Colombo, Country Director Italy of Meta, artificial intelligence in Europe “struggles to move”. “We,” he explained, “invest between 15 and 20 billion dollars in research and development per year that are not put to use when they arrive in Europe. There is a need to improve on the regulatory side” because “the EU may not have developed the large models that are the basis of this revolution today, but it has a lot of room when building applications”. Meta is working with the European Commission to “find a path that continues to guarantee value, but the fragmentation or lack of clarity of the direction to take does not help anyone”, added Colombo; “we have already had to change our strategies twice in the last year to meet the new European rules, in particular on personalized advertising, which is essential for companies, especially for small and medium-sized companies that are the backbone of our economic system. It is time to take stock of the situation because this uncertainty does not help us, nor the rest of the market”.