Detailed overview
Italy regulates AI through the EU AI Act and through its own national AI law, Law No. 132 of 23 September 2025 on artificial intelligence. The Italian law entered into force in October 2025 and sets national rules for AI research, development, experimentation, adoption and use. It operates alongside the EU AI Act and does not replace the EU framework.
Human-centred framework
The Italian AI law uses a human-centred model. AI must be developed and used in accordance with fundamental rights, EU law, transparency, proportionality, security, personal-data protection, confidentiality, accuracy, non-discrimination, gender equality and sustainability. The law also requires respect for human autonomy and human decision-making, and it refers to explainability, human oversight and human intervention.
Italy's law confirms that it does not create additional obligations beyond the EU AI Act for AI systems and general-purpose AI models where the EU AI Act already applies. This means that high-risk AI, prohibited AI and general-purpose AI obligations remain primarily governed by the EU AI Act. Italy's national law adds local principles, sector-specific rules and an authority structure.
Sector rules
In healthcare, AI may support prevention, diagnosis and treatment, but it must not be used to select or condition access to healthcare through discriminatory criteria. Patients must be informed when AI is used. AI can support healthcare professionals, but the final medical decision remains with the doctor or healthcare professional. Systems must be reliable, verified and periodically updated to reduce errors.
In employment, AI must be safe, reliable and transparent. Its use must be compatible with human dignity and privacy. Employers must inform workers when AI is used for automated decision-making or monitoring. The information must cover the aspects of the employment relationship affected by the system, the purposes and logic of the system, the data categories and parameters used for programming or training, control measures, accuracy, robustness, cybersecurity and possible discriminatory effects. Workers may request additional information, and workers and trade-union representatives must be informed about relevant AI use and significant changes.
For intellectual professions, AI may be used only as an instrumental support tool. The human professional's intellectual work must remain predominant, and the client must be informed clearly and completely when AI is used. This rule is relevant to lawyers, consultants, accountants, engineers and other regulated or intellectual professions.
In the justice system, AI may support administrative and analytical functions, but the interpretation of law, assessment of facts and evidence, and adoption of judicial measures remain reserved to the magistrate.
Authorities
Italy designates AgID and the National Cybersecurity Agency, ACN, as key AI authorities. AgID is responsible for innovation-related functions and certain notification and conformity-assessment matters. ACN has cybersecurity, market-surveillance, inspection and sanctioning functions. The Italian data-protection authority, communications authority and financial-sector regulators retain their existing powers where relevant.
Penalties
Penalties under the EU AI Act apply where the AI Act is breached. Italy's law also delegates the Government to adapt national legislation to the EU AI Act and provide the necessary supervisory, inspection and sanctioning powers.
Italian criminal law is also amended for certain AI-related offences. Unlawful dissemination of AI-generated or AI-altered images, videos or voices without consent, where this causes unjust damage, may be punished with imprisonment from one to five years.
AI may also operate as an aggravating circumstance where it is used to commit certain offences in an especially insidious way or to obstruct defence.