Artificial intelligence has entered a new phase that can be defined as mature rather than experimental: a transition that marks the beginning of an era focused on measuring real impacts and defining regulations. In this new scenario, competitive advantage does not stem from the mere possession of AI tools, but from the ability to govern and direct them—a competence that has taken on a political dimension, alongside industrial and infrastructural ones, shifting global competition to the realm of technological sovereignty.
The data presented in the third edition of the Annual Report by the Permanent Observatory on the Adoption and Integration of Artificial Intelligence (IA2) indicate that the challenge for organizations and institutions is no longer the availability of technology, but rather its effective integration into processes. In 2025, the adoption of artificial intelligence in Italy shows a marked size-based gap: 53.1% of large companies used at least one AI technology, compared to just 15.7% of SMEs. On the consumer front, only 19.9% of Italians claim to have used generative AI tools, a figure that is likely underestimated. Furthermore, 58% of companies that evaluated AI solutions ultimately did not adopt them due to skills shortages and the sheer scale of the required organizational transformation. Indeed, this transformation must be built upon a redesign of workflows and managerial roles, while productivity metrics are bound to shift from the individual to the organizational level.
The ability to lead rather than endure the transition is central to industry. Today, Italian manufacturing exports are exposed to geopolitical changes and the regionalization of markets, necessitating an outsourcing strategy based on value-added services. To this end, an adequate industrial policy is required—one committed to funding transformation platforms that involve both leading companies and widespread innovation.
Changes are also affecting the banking and financial sector, where the transition from adopting to implementing AI turns this technology into a customer-centric lever for transformation: structured listening is favored, and priority is given to the human element and guided experimentation. A similar strategy can be adopted by the public administration, a sector where the adoption of artificial intelligence must be geared toward strengthening effectiveness, trust, and proximity to citizens. These processes require proportionate, effective, and non-bureaucratic rules capable of guiding a responsible use of technology without hindering innovation.
Equally crucial is the development of human capital: the renewed centrality of soft skills—critical thinking, creativity, empathy—must be combined with the promotion of a widespread digital culture, an aspect far more strategic than mere technical skills. Moreover, a recurring fact emerges from the growing experience of Italian companies with AI: while technology proves to be a valuable tool, human oversight remains indispensable to avoid the risks of over-delegating to automated systems. For this reason, it is necessary to reaffirm the centrality of the ethical dimension alongside the legal one, especially for long-term issues that the law, by its nature oriented toward the present, struggles to fully protect.
Ethics and digital culture take on particular relevance in the relationship between AI and information. Several phenomena deserve attention to protect pluralism, as well as the “cognitive freedom” of citizens: the crisis of traditional media is accompanied by a growing disintermediation of information, which is accelerated by answer engines—AI-driven search tools that replace traditional search engines. To address these issues, we must start from the available regulatory instruments—from the Digital Services Act to the Italian legislation on fair compensation for publishers, recently upheld by a ruling of the Court of Justice—ultimately devising and implementing more effective methods to guarantee transparency in relations between platforms and publishers.
However, the challenge of protecting pluralism must not overshadow the global geo-economic framework. Europe risks a structural dependence on foreign suppliers, with direct consequences for its strategic autonomy. For this reason, it is called upon to consider creating continent-scale centers of excellence for AI: establishing an artificial intelligence structure modeled after CERN would help close the gap with China and the United States, promoting basic research and supporting the economic fabric in applied research.
The European delay is already considerable. On a macroeconomic level, AI-related growth is highly unequal globally: while it stands between 15% and 25% in the United States and between 10% and 15% in China, in Europe it hovers around 5%. Italy presents even lower estimates, ranging between 1% and 3%. Only by moving from adoption to a true integration of AI into collective processes will the country be able to recover lost ground and build a distinctive trajectory, founded on its own excellence and on the ability of institutions to lead—rather than endure—the transformation.