On 10 June 2026, the Italian Council of Ministers gave preliminary approval to the implementing legislative decrees of Law No. 132/2025, defining the country’s first comprehensive national regulatory framework on artificial intelligence. Italy is thus aligning with the European AI Act (EU Regulation 2024/1689), adopting an anthropocentric approach that places fundamental rights and human dignity at its centre.
Governance: Who Controls AI in Italy
The institutional architecture establishes a clear division of responsibilities. The Agency for Digital Italy (AgID) takes on the role of notifying authority, while the National Cybersecurity Agency (ACN) becomes the market surveillance authority and single point of contact with the European Union. Joining them are sector-specific bodies such as the Bank of Italy, Consob and Ivass for the financial sector, and the Data Protection Authority for high-risk applications in the judicial and security fields.
Training: €100 Million for Schools and the Public Administration
Training is the central pillar of Italy’s strategy. In schools, AI must be permanently integrated into civic education and upper secondary curricula, backed by a dedicated fund of €100 million earmarked for teacher training. Universities are introducing compulsory interdisciplinary laboratories combining technical, legal and ethical profiles.
For the Public Administration, three-tier programmes are planned: basic literacy for all staff, specialist retraining for administrative procedures, and advanced training for executives responsible for digital transition.
Employment: Automated Decisions on Workers Banned
On the employment front, the decree establishes a fundamental principle: decisions relating to the formation, modification or termination of the employment relationship — including dismissals and disciplinary measures — cannot be taken in a solely automated manner. Any decision affecting a worker’s rights must be reserved to a human being. A dismissal decided by an algorithm alone is null and void by law.
Before any algorithmic processing, the employer must fulfil precise disclosure obligations. The worker has the right to obtain, through the intervention of a human operator, a statement explaining the parameters considered by the AI system and its weight in the decision-making process.
Justice and Security: Clear Limits on AI Use
In the area of policing, the rules prohibit mass surveillance and the creation of databases through indiscriminate scraping of the web. Real-time biometric recognition is permitted only for serious threats to security, terrorism or the search for missing persons, subject to authorisation from the judicial authority for a maximum of 15 days.
On the criminal side, Article 437-bis of the Penal Code is introduced, punishing the failure to adopt or the alteration of security measures in high-risk AI systems, and extending administrative liability to legal persons under Legislative Decree No. 231/2001.
Investment: Up to One Billion for Startups
To support the industrial ecosystem and the country’s digital sovereignty, Article 23 of Law No. 132/2025 allocates up to one billion euros from the Venture Capital Support Fund. More than €300 million have already been allocated by CDP Venture Capital to over 150 startups, with €500 million in investments planned over the next three years and the launch, from 2026, of the SophIA Hub with an allocation of approximately €30 million.
Original article: Fiscoetasse.com
