Artificial Intelligence: the Council of Ministers has approved the Implementing Decrees
On June 10, 2026, the Council of Ministers approved two schemes of implementing decrees pursuant to Law No. 132/2025, entitled “Provisions and Delegated Powers to the Government Regarding Artificial Intelligence.”
The main topics covered are listed below:
A) Scheme of implementing decree regarding the National Authorities' powers and the use of artificial intelligence in learning
The decree provides guidelines on training in the use of artificial intelligence systems, intended to increase awareness and understanding of the risks associated with the use of such tools. Specifically, the decree sets out rules for the use of AI in school and university programs, within public administrations, in the healthcare sector, and in professional practice.
The second issue concerns the protection of workers and, specifically, data processing procedures and ensuring that decision-making is not based solely on outputs generated by artificial intelligence.
In terms of the administration of justice, particular attention is reserved for training programs, given that the AI Act has identified the judicial system as a high-risk sector.
Finally, the draft decree identifies national authorities involved in the implementation of the AI Act, namely (i) AgID, (ii) ACN, (iii) Banca d’Italia, CONSOB, and IVASS, and (iv) the Italian Data Protection Authority.
B) Scheme of implementing decree on the use of artificial intelligence systems in police operations and on civil and criminal liability
The second implementing decree concerns the administration of justice.
Firstly, the decree aims to regulate - including through amendments to the Code of Criminal Procedure - the cases and procedures for using artificial intelligence tools for biometric identification, in order to prevent and fight crime.
Secondly, it regulates civil liability for damages caused by artificial intelligence tools.
The most significant issue concerns the introduction of a new criminal offense aimed at punishing the failure to implement security measures in high-risk artificial intelligence systems and the manipulation of existing security measures, where such omissions or acts result in a concrete danger to life, public safety, or national security. Such conduct, in addition to being punishable as a intentional offence, would also constitute a crime in cases of gross negligence.
Furthermore, the new offence would also constitute a predicate offense for administrative liability of legal entities, which would require an update to the risk assessment and the Organizational, Management, and Control Models.