Facial Recognition Yes, But Only This Way: The Rules of Meloni's Government to Tame Artificial Intelligence
The Council of Ministers has approved two legislative decrees for preliminary examination implementing law n. 132/2025, completing the first comprehensive national regulatory framework on artificial intelligence in Italy. The measure is fully in line with the European AI Act (EU Regulation 2024/1689) and does not introduce an alternative regulation but ensures its implementation in the domestic legal system. Before coming into force, the texts will be subjected to scrutiny by parliamentary committees, the Conference of Regions, and the competent authorities.
The first decree intervenes extensively in the educational system. In secondary schools, national guidelines are being updated to integrate generative AI into study paths, with the introduction of the subject within civic education and the strengthening of STEAM (Science, Technology, Engineering, Arts, and Mathematics) skills. A training plan for teachers is foreseen, with 100 million euros allocated, explicitly aimed at combating digital addiction and algorithmic opacity among minors. Universities, AFAM institutions, and ITS Academy will need to integrate content on the safe and informed use of AI systems, with an interdisciplinary approach that includes technical, legal, and ethical profiles. The monitoring of educational quality is entrusted to ANVUR.
For the public administration, the decree provides for three levels of intervention: basic literacy for all employees, professional retraining for specific functions, and advanced training for executives. The National School of Administration (SNA) will translate these guidelines into homogeneous modular paths. In the health sector, mandatory training on AI is included in the ECM program with a dedicated percentage, also extended to health executives for managing waiting lists and optimizing resources. Professional orders have six months to adapt their regulations, with pathways covering technical, legal, and ethical aspects. The professional's fee must reflect the risk level of the AI system used, with tariff parameters to be updated within twelve months.
On the labor front, the decree establishes that decisions regarding the establishment, modification, or termination of employment relationships - including disciplinary measures and dismissals - cannot be made solely based on automated treatments. The final decision rests with a natural person. The worker has the right to receive understandable reasoning, upon request, indicating the possible impact of the AI system on the decision-making process. Dismissals carried out in violation of this rule are null and void.
The second decree regulates the use of AI in police activities. Remote real-time biometric identification is allowed exclusively to prevent specific threats to public safety or for the search of missing persons and victims of serious crimes, subject to judicial authorization valid for a maximum of fifteen days, renewable with justification. Post-facto facial recognition is allowed only after a crime has been committed, based on video-photographic documentation, with logs preserved for five years. The establishment of biometric databases through mass and non-targeted data collection from the web is prohibited.
On the civil front, the decree aims to rebalance the position of those who suffer damage from an AI system, providing access to the technical documentation of the system, a presumption of causal link, and the possibility to act directly against the insurance. On the criminal front, Article 437-bis is introduced into the penal code, sanctioning the failure to adopt safety measures in high-risk AI systems and their alteration when this leads to a concrete danger for life, public safety, or state security. Responsibility may extend to entities under D.lgs. 231/2001.
AgID takes on the role of notification authority, while ACN is designated as the market supervisory authority and point of contact with EU institutions. Bank of Italy, CONSOB, IVASS, and the Privacy Guard exercise specific sector functions. On the industrial front, law 132/2025 allocates up to one billion euros from the venture capital support fund to the development of the national AI ecosystem. The Italian market has reached 1.8 billion euros in 2025, growing by 50% year-on-year. From 2026, the SophIA hub, dedicated to AI and cybersecurity, will be operational, with an initial endowment of approximately 30 million euros.