Written by: Haim Ravia, Dotan Hammer
On September 17, 2025, the Italian Senate approved a law on Provisions and Delegations to the Government Regarding Artificial Intelligence. The legislation promotes the correct, transparent, and responsible use of AI in Italy, prioritizing the human dimension. It aims to ensure vigilance over the economic and social risks associated with AI and its impact on fundamental rights. The provisions are to be interpreted and applied in conformity with the EU’s AI Act.
Key operational principles for AI development include respecting fundamental rights, EU law, and ensuring transparency, proportionality, safety, non-discrimination, and personal data protection. The law emphasizes that AI systems must be developed and applied while respecting human autonomy and decision-making power, ensuring human surveillance and intervention. Cyber-safety is defined as an essential precondition throughout the AI life cycle.
The bill establishes a regulatory structure by designating two National Authorities for AI: the Agency for Digital Italy (AgID) and the National Cybersecurity Agency (ACN). AgID is responsible for promoting innovation and development, while ACN handles surveillance, inspection, and sanctions, particularly concerning cybersecurity.
The law also governs the use of AI in specific sectors:
- In healthcare, AI systems serve as support, but the final decision is always reserved for the medical professional. The introduction of AI cannot select or condition access to healthcare services based on discriminatory criteria.
- In the workplace, AI must be safe, reliable, and transparent, and its use cannot conflict with human dignity or violate personal data confidentiality. The employer must inform the worker about the use of AI.
- In justice and public administration, AI is used as an instrumental and supporting function; however, the human (magistrate or administrator) remains solely responsible for the final decisions and proceedings.
The law also includes provisions authorizing up to one billion Euros to support innovative small and medium enterprises (SMEs) operating in the sectors of AI, cybersecurity, and enabling technologies. Furthermore, the law introduces modifications to the penal code, including new aggravating circumstances for crimes committed using AI systems, and establishes sentences (imprisonment from one to five years) for the illicit diffusion of images, videos, or voices falsified or altered by AI (deepfakes) that are misleading and cause unjust damage.
Click here to read Italy’s new AI law (in Italian).
