On September 17, 2025, the Senate definitively approved the new Italian law on artificial intelligence (“AI”), which will come into force after its publication in the Official Gazette.
This law (hereinafter referred to as the “Italian AI Law”) establishes a series of fundamental principles relating to research, experimentation, development, and adoption of artificial intelligence systems with “correct, transparent, and responsible use, in an anthropocentric dimension.” These principles include respect for human autonomy and decision-making power, the permanence of human oversight and intervention, and the ultimate responsibility of the human being using AI (public official, magistrate, professional), and sets limits on such use related to the protection of the constitutional rights of the individuals concerned.
Since the subject of AI has become a competence of the EU following the enactment of the recent Regulation (EU) 2024/1689 (known as the “AI Act”), which by its nature is directly applicable in Member States, the Italian AI Law cannot introduce requirements that are different or incompatible with the AI Act (and in fact, the Italian law expressly states that it does not establish additional obligations with respect to the AI Act), but it can, as it has done, introduce general principles, define the powers of administrative authorities, and delegate the Government to ensure the complete and orderly implementation in Italy of the AI Act and the provisions of the Italian AI Law itself.
However, it would be reductive to consider the Italian AI Law merely as an act of adapting the national legal system to the AI Act, as it also considers the strategic implications and effects of AI on democratic debate, establishing that the use of AI ‘must not prejudice the democratic functioning of institutional and political life and the exercise of territorial powers based on the principles of autonomy and subsidiarity, nor must it prejudice the freedom of democratic debate from unlawful interference, caused by anyone, protecting the interests of state sovereignty and the fundamental rights of every citizen recognized by national and European law’. Here, the legislator has certainly had in mind the “fake news” created with AI to influence public opinion on current hotbeds of tension in the world.
On an economic level, the Italian AI Law recognizes the opportunities arising from the use of artificial intelligence, such as technological independence and increased productivity for national companies, while also taking into account, with support and training provisions, the Italian economic fabric made up of small and medium-sized enterprises.
From the point of view of the institutions responsible for AI supervision and guidance, the Italian AI Law establishes that the Presidency of the Council (through the Department for Digital Transformation) will develop a National Strategy to be updated periodically, while the national authorities for AI will be the Agenzia per l’Italia digitale (Agency for Digital Italy or AgId) and the Agency for National Cybersecurity (Acn), without prejudice to the specific competences of the Bank of Italy, Consob, Ivass (insurance) and Agcom (competition, market and consumer protection). In particular, AgId is responsible for promoting innovation and the development of artificial intelligence, while Acn, also with a view to ensuring cybersecurity, is responsible for inspection and supervision and for imposing sanctions in relation to artificial intelligence systems. Representatives of the Bank of Italy, Consob, and Ivass participate in the meetings of the AI Coordination Committee, composed of the heads of the Department for Digital Transformation of the Presidency of the Council of Ministers and AgId and Acn, when such meetings concern matters falling within the competence of one or more of these three market supervisory authorities.
The Italian AI Law consists of six chapters, of which Chapter I summarises the principles and objectives described above.
Chapter II contains sector-specific provisions covering the use of AI in healthcare—as a support for treatment and diagnosis, but with the final decision remaining with doctors—and in scientific research, mandating the Minister of Health to coordinate the processing of certain personal data with AI in accordance with the provisions of the GDPR, simplifying procedures as much as possible. Chapter II also regulates the use of AI in the workplace and in the professions, requiring employers and professionals to inform employees or clients, respectively, if AI is used, as well as in public administration, again with an anthropocentric approach, with the use of AI limited to ‘instrumental and support activities’. The same principles underlie the provisions of Chapter II on the use of AI in the judiciary, where ‘all decisions on the interpretation and application of the law, the assessment of facts and evidence, and the adoption of measures are always reserved to the judge’, meaning that a judicial decision generated by AI is never possible. Chapter II also contains a delegation to the Government to issue decrees to provide comprehensive regulation of the use of data, algorithms, and mathematical methods for training artificial intelligence systems and, finally, provides for the use of AI, including through agreements with private parties, to “enhance artificial intelligence as a resource for strengthening national cybersecurity.”
Chapter III covers, among other things, the development of the national strategy for artificial intelligence, and contains the delegation of powers to the Government to issue decrees to implement the provisions of the AI Act into the Italian legal order, expressly providing, among other things, for amendments to the rules on banking, financial, payment, and insurance services, and the “definition of the supervisory powers of the market supervisory authority, granting the authority the powers to require providers and potential providers to transmit information, to carry out remote or on-site inspections, even without prior notice, and to carry out checks on the conduct of tests in real conditions and on the related high-risk artificial intelligence systems.”
Finally, Chapter IV contains provisions on copyright, recognizing a certain degree of protection for intellectual works created with AI, Chapter V contains criminal provisions introducing aggravating circumstances if certain crimes are committed using AI, and Chapter VI contains the final provisions.
The Italian AI Law is certainly interesting so long as it introduces general criteria regarding the purposes, methods, and limits of AI use, which economic operators and supervisory authorities will have to refer to when assessing relevant cases. Furthermore, the delegated powers contained in the same Law suggest that rules on the use of AI will also be issued in financial intermediation, which investment service providers and fund managers will have to comply with when using AI in services to Italian customers.
Recent Comments