Italy has positioned itself as the first European Union member state to adopt a comprehensive national framework on the use of artificial intelligence, through Law no. 132/2025 (the National AI Law) adopted on October 10, 2025. The National AI Law is intended to complement Regulation (EU) 2024/1689 (the EU AI Act) by setting sector specific rules that fall under national competence and establishing the institutional architecture for artificial intelligence oversight in Italy, including designated authorities and administrative and criminal sanctions. Unlike the EU AI Act, which provides for gradual implementation, several provisions of the National AI Law are already in force and fully applicable, while the government is mandated to adopt implementing decrees within twelve months, i.e., by October 10, 2026, on issues such as training of artificial intelligence systems, civil redress, alignment with the EU AI Act, civil procedure rules on damages, and criminal sanctions.
The National AI Law adopts the definitions of “Artificial Intelligence” and “AI system” contained in the EU AI Act and underlines an anthropocentric approach that requires research, development and deployment of artificial intelligence systems and models to respect human rights and constitutional and EU law principles. Its territorial scope covers the development and deployment of artificial intelligence systems and models in Italy and it operates alongside other relevant frameworks, including the GDPR, the Italian Data Privacy Code, competition rules, consumer law, and intellectual property law such as the Italian Copyright Law and the Industrial Property Code. The Italian Data Protection Authority has issued a notice on the use of web scraping for training artificial intelligence models and created a dedicated artificial intelligence unit, while courts, including the Italian Supreme Court, have begun applying existing laws to artificial intelligence use cases, particularly in relation to biometric data processing. The National AI Strategy further frames policy around strengthening expertise and talent, increasing funding for advanced research in artificial intelligence, and encouraging artificial intelligence adoption across public administration and productive sectors.
Sector specific provisions focus on healthcare, labor, professional services, public administration, and copyright. In the national health service, artificial intelligence may support disease prevention, diagnosis and treatment but cannot be used to discriminate in access to care, must promote inclusion for persons with disabilities, and may only support medical procedures, leaving final decisions and liability with medical professionals, who must inform patients when artificial intelligence is used. Public and non-profit entities may process personal data for artificial intelligence development as a matter of “paramount public interest” with general notice, anonymization, and prior notification to the Data Protection Authority at least 30 days in advance, and the Minister of Health is expected to issue guidelines by February 2026 and oversee a national artificial intelligence platform to assist professionals and patients. In the workplace, artificial intelligence use must be safe, reliable and transparent, with mandatory information to employees and an Observatory established to monitor labor market impacts and promote training. For intellectual professions, artificial intelligence systems may only support professionals, who must ensure the prevalence of intellectual work and inform employers or clients when artificial intelligence has been used. Public bodies and the judiciary may use artificial intelligence solely to support decisions, with responsibility retained by officials and judges, and the Minister of Justice tasked with regulating use and training. The law clarifies that copyright protects only works of human intellect, including those created with artificial intelligence where there is human creative input, excludes protection for content generated solely by machines, expressly regulates text and data mining by artificial intelligence subject to existing exceptions, and criminalizes unauthorized text and data mining and scraping.
The compliance and enforcement architecture aligns with the EU AI Act without introducing a separate risk categorization, while designating national authorities to oversee artificial intelligence. The Agency for Digital Italy (AgID) will act as the notifying authority for accreditation and supervision of conformity assessment entities and will promote innovation and development of artificial intelligence, whereas the National Cybersecurity Agency (ACN) will be the supervisory and lead authority for artificial intelligence in cybersecurity, without prejudice to the sectoral powers of the Bank of Italy, CONSOB, IVASS, the Data Protection Authority and AGCOM. The National AI Law introduces a specific criminal offense of “Unlawful dissemination of content generated or altered with AI systems” in Article 612-quater of the Penal Code, targeting deepfakes where artificial intelligence altered content likely to mislead and cause unjust harm can result in one to five years’ imprisonment, and provides a new aggravating circumstance that increases sentences by up to one third when any crime is committed using artificial intelligence or when artificial intelligence is used to hinder defense or aggravate consequences. Additional criminal penalties apply to crimes such as attacks on political rights, market rigging and market manipulation when committed with artificial intelligence, and copyright sanctions are extended to unauthorized scraping and abusive text and data mining. The government must issue further implementing decrees within twelve months to define sanctioning powers and additional offenses, govern use of data, algorithms and mathematical methods for training artificial intelligence systems, and establish civil redress mechanisms and related sanctions, while disputes on the functioning of artificial intelligence systems are assigned to specialized business sections of civil courts through amendments to the Code of Civil Procedure.
