On June 25, 2025, the Italian Chamber of Deputies approved, in its second reading, draft law no. 2316-A entitled “Provisions and delegations to the Government on the subject of artificial intelligence” (hereinafter, the AI Bill). With 136 votes in favor, 94 against, and 5 abstentions, the bill—previously approved by the Senate on first reading on March 20—is now awaiting a third reading in the Senate due to amendments made since the original version.
The AI Bill aims to complement and support the entry into force of Regulation (EU) 2024/1689 on artificial intelligence (AI Act) by establishing general principles for the use of AI systems in the Italian context and by delegating to the Government the adoption of legislative decrees to regulate operational aspects, with particular attention to transparency, safety, accountability, and data use.
Coordination with the European AI Act
The parliamentary process of the AI Bill was marked by the need for ongoing coordination with Regulation (EU) 2024/1689, which came into force on August 1, 2024. The bill’s provisions are to be interpreted and applied in accordance with the AI Act, as established in Article 1, paragraph 2. This approach is particularly significant: Italy is positioning itself to become the first EU country to adopt a comprehensive national law on artificial intelligence, in implementation and integration of the European regulation.
The European Commission initially identified misalignments and conflicts between national and EU provisions, as outlined in opinion C (2024) 7814 of November 5, 2024. The current version of the bill has incorporated these comments to avoid inconsistencies that could lead to the national rules being disregarded. The goal is to create a regulatory framework that, while respecting the principles of the AI Act, also considers the specificities of the Italian legal system and the needs of the national production sector.
A human-centered and sectoral approach
The AI Bill adopts a human-centered approach aimed at preserving human centrality and oversight in decision-making processes involving AI systems. This principle, aligned with the AI Act, is expressed in Article 3, which states that research, experimentation, development, adoption, application, and use of AI systems and models must respect fundamental rights and freedoms under the Italian Constitution, EU law, and principles such as transparency, proportionality, safety, personal data protection, privacy, accuracy, non-discrimination, gender equality, and sustainability.
The bill also takes a sectoral approach, dedicating specific provisions to areas considered high-risk under the AI Act: justice, healthcare, labor, public administration, and regulated professions. This legal strategy aims to provide precise, contextualized rules for sectors that have significant impacts on fundamental rights and social organization.
Amendments introduced by the Chamber: stronger protection for minors
The Chamber of Deputies introduced limited but substantial amendments, necessitating the bill’s return to the Senate for a third reading. The most significant change concerns Article 4, paragraph 4, where the phrase “as well as the consequent processing of personal data requires” was added. This broadens the scope of parental consent requirements for minors under fourteen.
This clarification is practically relevant, as the use of AI systems inevitably involves collecting, processing, and storing minors’ personal data—often in ways that are complex and not immediately apparent to the user. The amendment therefore extends parental consent not only to access AI technologies but also to all related data processing activities, significantly strengthening protections for vulnerable individuals.
Sector-specific rules: from justice to healthcare
In the justice sector, the AI Bill introduces a particularly detailed framework. Article 15 reserves to judges all decisions regarding the interpretation and application of the law, the evaluation of facts and evidence, and the issuing of rulings, expressly prohibiting the use of AI for interpretive or decision-making activities. However, AI may still be used in support of administrative, organizational, or predictive analysis tasks, provided that human oversight is maintained.
Noteworthy is also Article 17, which grants exclusive jurisdiction to the courts for cases concerning the operation of an AI system, introducing a significant innovation in the Italian judicial system.
In healthcare, the bill establishes that AI use must not result in discrimination and that decisions concerning prevention, diagnosis, treatment, and therapeutic choices must always be made by medical professionals. AI may support the improvement of disease prevention, diagnosis, and treatment, but only as a support tool. Human oversight must be maintained in clinical decisions, and patients must be informed about the use of such technologies.
Governance and competent authorities
The bill establishes a complex governance system, designating the Agency for Digital Italy (AgID) as the authority responsible for notification, evaluation, accreditation, and monitoring of the entities tasked with conformity assessments. The National Cybersecurity Agency (ACN) is assigned supervisory authority. The specific responsibilities of the Bank of Italy, CONSOB, and IVASS remain intact in their respective areas of market supervision, as do those of the Italian Data Protection Authority.
Additionally, a national strategy on AI is to be adopted and updated every two years by the Interministerial Committee for Digital Transition, supported by the Department for Digital Transformation. The aim is to coordinate public policies on AI and promote technological innovation while respecting fundamental rights.
Implementation implications and operational recommendations
The ongoing regulatory evolution requires organizations—both public and private—to take proactive steps now to ensure compliance with the upcoming legal framework. Specifically, companies and public administrations will need to immediately audit the AI systems currently in use, assess the legal basis for their deployment, verify the level of transparency provided to users, and evaluate associated risks.
Particular attention must be paid to age verification mechanisms for services directed at minors. The amendments introduced by the Chamber make it essential to implement robust systems for obtaining informed parental consent, not only for access to AI technologies but also for all subsequent personal data processing.
From a governance perspective, it is recommended to establish dedicated ethics committees and strengthen dialogue among corporate functions: Data Protection Officers, HR, compliance, and technical teams must work together to ensure an integrated risk management approach. Equally crucial will be the revision of contracts with AI technology providers and cloud platforms, considering new requirements related to digital sovereignty and transparency.
Finally, organizations should prepare detailed documentation of decision-making processes involving AI systems, ensuring compliance now with the transparency and accountability standards that will be mandated by forthcoming delegated decrees. This proactive approach will not only help anticipate future legal obligations but also build competitive advantage through user trust and robust technological implementation.
Outlook and final considerations
The approval of the AI Bill represents a major step toward establishing a comprehensive regulatory framework for AI in Italy. It fits within an evolving European context, where the AI Act sets harmonized principles and obligations but allows Member States to regulate specific and sectoral aspects.
Although limited in number, the amendments introduced by the Chamber demonstrate lawmakers’ focus on protecting the most vulnerable and the need to strike a balance between innovation and rights protection. The return to the Senate for a third reading marks the final phase of a complex legislative process that has involved numerous stakeholders through hearings and consultations.
The effectiveness of the bill will ultimately depend on the adoption of implementing legislative decrees by the Government, which must occur within one year from the law’s entry into force. It will therefore be essential to closely monitor the development of secondary legislation, which will translate the bill’s principles and delegations into operational rules.
