Sardinia Regional Law, March 10, 2026 — Provisions on Promotion, Sustainable Development and AI Governance in Sardinia

1. Introduction and Systematic Framework

With Regional Law of March 10, 2026 (hereinafter "the Law"), the Regional Council of Sardinia adopted the first comprehensive sub-national legislative framework in Italy entirely dedicated to the promotion, sustainable development and governance of artificial intelligence. The act sits within a layered regulatory context — EU Regulation 2024/1689 ("AI Act") and national transposition legislation — and, respecting the constitutional allocation of competences, covers typically regional areas of intervention: productive system innovation, training, health, territorial governance and administrative organization.

The Law comprises 22 articles and pursues four stated objectives (Art. 1(1)): fostering the economic and social progress of the territory; ensuring citizen well-being through innovative public services; accelerating responsible AI adoption in strategic sectors; preparing the workforce for the opportunities of the digital era. The legislative technique adopted is that of framework laws: the Law sets objectives, principles and governance structures, deferring implementing regulation to the Regional Government (Giunta regionale) by resolution.

2. Relationship with the AI Act and Higher-Level Regulatory Framework

2.1 Vertical Regulatory Allocation

The AI Act (EU Reg. 2024/1689), entered into force on August 1, 2024 and progressively applicable until August 2, 2026, is an EU secondary law instrument directly applicable in all Member States. It regulates the placing on the market, putting into service and use of AI systems, adopting a risk-based approach and classifying systems into four categories: prohibited, high-risk, limited risk, and minimal risk.

The Regional Law self-qualifies as legislation adopted "in compliance with relevant EU and national legislation" (Art. 1(1)). The reference to the definitions of EU Reg. 2024/1689 (Art. 2(1)(a) and (2)) implements a formal "receptive" referral technique that ensures dynamic terminological consistency: whenever the EU legislator updates the notion of "artificial intelligence system", the regional definition updates automatically.

2.2 Regional Areas of Intervention

EU Reg. 2024/1689 does not exhaust the AI field: it leaves national and sub-national legislators wide margins on promotion, institutional governance, training and public procurement. The Law operates predominantly in these residual and complementary spaces:

• Institutional governance (Arts. 3-8): establishment of structures, Hub and Observatory, a matter not harmonized by the AI Act.
• AI use in regional and local PAs (Arts. 9-10): the AI Act sets minimum requirements for high-risk systems used by public authorities; the Regional Law supplements these with organizational and training principles.
• Training and research (Arts. 13-15, 18): concurrent or residual regional competence.
• Healthcare (Art. 16): concurrent State-Region competence (Art. 117(3) Constitution).
• Business support (Art. 17): concurrent competence, with State aid constraints (Art. 19).

3. The Regional AI Governance System

3.1 Institutional Architecture

The Law designs a multi-level governance system structured on three tiers:

• (i) Regional administrative structure (Art. 3(1)): identified by Government resolution on the President's proposal, it is the hub for programming, coordination, implementation and monitoring of regional AI policies.
• (ii) HUBIAS – Regional AI Hub (Art. 5): CRS4 is designated as the Hub's seat, with tasks of advanced research, applied experimentation, validation and prototyping. CRS4 also coordinates territorial open innovation hubs (Art. 8). It is the technical-scientific core of the system.
• (iii) Regional AI Observatory (Art. 6): a collegial body with plural composition (Government, departments, universities, local authorities, social partners), established at no additional cost to the regional budget.

3.2 The Digital Transition Officer (RTD)

The Law recognizes a liaison role for the RTD, a figure established by Art. 17 of the Digital Administration Code (Legislative Decree 82/2005). The administrative structure under Art. 3 operates "in close collaboration and coordination" with the RTD: this avoids duplication and ensures continuity with the broader digital agenda of the Regional Administration.

3.3 The AI Systems Registry (Art. 12)

The Law establishes a public and online-accessible registry of AI systems used by regional and local public bodies, at no additional budget cost. The establishment is consistent with the principle of algorithmic transparency enshrined in the AI Act and anticipates, at the regional level, transparency obligations that the AI Act already imposes on high-risk system operators.

4. Artificial Intelligence in Public Administration

4.1 Foundational Principles (Art. 9)

Art. 9 codifies a catalogue of principles for AI adoption in public administration, largely derived from the AI Act and the Ethics Guidelines for Trustworthy AI by the HLEG (2019):

• Algorithmic transparency and decision explainability
• Human oversight
• Accountability
• Fairness and non-discrimination
• Security and resilience
• Personal data protection and confidentiality (implicit reference to GDPR, EU Reg. 2016/679)
• Sustainability, inclusivity and accessibility

4.2 Cybersecurity (Art. 11)

Art. 11 requires the Region, regional bodies, local authorities and in-house companies to adopt cybersecurity measures compliant with the NIS 2 Directive (2022/2555/EU) and ACN (National Cybersecurity Agency) guidelines. Notably, provinces are called upon to support municipalities in their territory, with priority to small municipalities (Art. 11(3)), filling a structural gap in the security ecosystem for smaller entities. The provision is fully consistent with the DORA framework (EU Reg. 2022/2554) for the financial sector.

5. Training: A Multi-Level Integrated System

The Law structures training on four levels, from primary school to adult professional requalification:

6. AI in Healthcare

Art. 16 regulates AI use in healthcare with a prudential approach, consistent with the AI Act classification that considers AI systems intended to assist clinical decisions as high-risk (Annex III, point 5). The provision establishes two fundamental principles: (i) AI systems act as support tools and not as substitutes for medical judgment; (ii) their use occurs in compliance with clinical responsibility, safety, transparency and data protection.

From a legal standpoint, maintaining medical judgment as an essential step is relevant to the civil and criminal liability of healthcare professionals. Italian case law applies the Gelli-Bianco Law (Law No. 24 of March 8, 2017) also to algorithmic support contexts.

7. Business Support and State Aid

7.1 Incentive Measures (Art. 17)

The Law provides a range of measures supporting the business ecosystem: strengthening technology transfer centers, public-private partnership networks, AI solution implementation in SMEs, and support for innovative startups (creation, growth, capital market access and internationalization).

7.2 State Aid Constraint (Art. 19)

Art. 19 is a safeguard provision: implementing acts constituting State aid under Arts. 107-108 TFEU must be notified to the European Commission, unless they fall under the GBER (EU Reg. 651/2014) or the de minimis regime (EU Reg. 2023/2831, aggregate threshold €300,000 over three fiscal years).

8. Financial Provision and Coverage

Art. 21 designs a tripartite funding framework: NRRP resources (Missions 01 and 04), EU structural funds (ERDF/ESF+) and regional resources. An experimental expenditure of €1,200,000 per year for the 2026-2028 triennium is authorized, financed through a reduction in Mission 20 – Programme 03 (FNOL). Coverage is therefore integral and does not burden the regional deficit.

9. Critical Assessment: Strengths

• Systematic approach: the Law provides Sardinia for the first time with a comprehensive AI framework, overcoming fragmented previous sectoral interventions.
• EU law consistency: the receptive referral to the AI Act ensures regulatory alignment and reduces the risk of conflicts between regulatory levels.
• Inclusive approach: attention to gender gap reduction, small municipalities, inner areas and workers at risk of technological displacement demonstrates sensitivity to Sardinia's territorial specificities.
• Evaluation clause (Art. 20): the annual report obligation to the Regional Council by October 31 provides a political accountability safeguard and enables timely regulatory adjustments.
• Financial coverage: the financial provision (Art. 21) indicates integral coverage through internal transfers, without creating new regional debt.

10. Opportunities for Businesses, Professionals and Consultants

The Law opens multiple operational windows for Sardinia's economic and professional system:

• Innovative startups and SMEs: access to grants, guarantee funds and internationalization programs (Art. 17). Businesses should promptly map their AI systems against the AI Act risk classification, preparing for full application on August 2, 2026.
• Innovation consultants and lawyers: the Observatory (Art. 6) and digital platform (Art. 7) create demand for AI Act compliance services, gap analysis, DPIA, AI adoption contracts and strategic advisory for incentive access.
• DPOs and data protection officers: Registry implementation (Art. 12) and Art. 9 principles imply a review of processing records (Art. 30 GDPR) and likely new DPIAs for high-risk AI systems adopted by public administrations.
• Research centers and universities: CRS4 and Sardinia's two universities (Cagliari and Sassari) are privileged recipients of AI doctorate and research funding (Arts. 14, 18).
• Local authorities: the Law provides technical, organizational and training support from the Region (Art. 10) and from provinces toward smaller municipalities (Art. 11(3)).

11. Conclusions

Sardinia Regional Law of March 10, 2026 represents a regulatory act of significant systematic scope in the landscape of Italian regional innovation policies. The choice to adopt a multi-level governance approach — administrative structure, technical-scientific Hub, plural Observatory — is commendable and well aligned with the AI Act's institutional architecture.

The key implementation challenges will involve: timely adoption of implementing acts by the Regional Government; specification of the AI Systems Registry content; operational coordination between the Art. 3 structure, the RTD, the regional DPO and national authorities (ACN, Privacy Authority, future AI Act supervisory body); and the concrete delivery of business incentives in compliance with State aid rules.

For the professional system, the Law inaugurates a new season of demand for legal services and specialized consulting in AI compliance, data governance and access to support measures. The time window runs between the Law's entry into force (date of publication in the BURAS) and the AI Act's full operability (August 2026): a precious semester to structure the professional offering and position as reference interlocutors for public administrations and businesses in the territory.