Italy Law No. 132/2025 on Artificial Intelligence (Legge 23 settembre 2025, n. 132) — provisions explorer

AI in Employment

Art. 11 — workplace AI must be safe, reliable, transparent, non-discriminatory and not contrary to human dignity; employer must inform the worker of AI use (per Art. 1-bis D.Lgs. 152/1997). Art. 12 establishes a national Observatory on workplace AI.

Editor rationale: Dedicated operative article on AI in employment with a concrete employer information duty (anchored to D.Lgs. 152/1997, hence medium not high).

AI in Criminal Justice

Art. 15 — in judicial use of AI, decisions on legal interpretation/application, evaluation of facts and evidence, and adoption of measures are always reserved to the magistrate; AI limited to organisational/administrative support. Art. 24(2)(h) delegates a future regime for AI in policing.

Editor rationale: Operative reservation of judicial decision-making to the magistrate is the closest match; confidence low because the topic centres on predictive/risk-scoring law-enforcement tools, which the law leaves to a future delegated decree (Art. 24(2)(h)).

Training-Data Rights

Art. 25 (new Art. 70-septies l. 633/1941) permits text-and-data-mining reproductions/extractions for AI training from lawfully accessible material (per Arts. 70-ter/70-quater); Art. 16 delegates the Government to enact an organic regime on data, algorithms and mathematical methods for training AI.

Editor rationale: Art. 70-septies is operative now (a TDM exception governing training inputs); low confidence because the substantive training-data regime is largely deferred to a future delegated decree (Art. 16).

Deepfakes / Synthetic Content

Art. 26(1)(c) inserts new Criminal Code Art. 612-quater: illicit dissemination of AI-generated or altered images/video/voices, without consent, apt to deceive and causing unjust harm — 1 to 5 years' imprisonment (querela-based; ex officio in aggravated cases).

Editor rationale: Dedicated new criminal offence directly targeting AI-generated/altered deceptive media; medium because it is a misuse-criminalisation modality, narrower than a general disclosure/labelling regime.

Transparency Obligations

Multiple operative disclosure duties: Art. 4(3) clear-language information on AI data processing + right to object; Art. 7(3) patient information; Art. 11(2) worker notification; Art. 13(2) professional's duty to disclose AI use to the client.

Editor rationale: Several concrete information/disclosure obligations across sectors; medium because they are sector-anchored duties, not a single horizontal transparency/labelling regime (left to the EU AI Act).

Technological Sovereignty

Art. 5 — the State must promote AI to raise national competitiveness and the 'technological sovereignty of the Nation' (¶1(a)) and may steer public e-procurement to favour solutions localising strategic data and disaster-recovery/business-continuity in national data centres (¶1(d)).

Editor rationale: Explicit textual invocation of 'technological sovereignty' plus an operative data-localisation procurement lever; low because the data-centre preference is discretionary ('possano essere privilegiate'), not a hard mandate.

National Security Carveouts in AI Regulation

Art. 6 — activities for national-security purposes by the intelligence services, ACN cybersecurity/resilience, national-defence by the Armed Forces, and certain national-security policing are excluded from the law's scope (subject to fundamental-rights respect; further rules by regulation under l. 124/2007 art. 43).

Editor rationale: Express, operative scope exclusion for the national-security/defence/intelligence domain — a textbook national-security carve-out; medium because qualified by a fundamental-rights clause and downstream implementing regulations.

AI in Healthcare

Art. 7 — AI must not condition access to healthcare on discriminatory criteria (¶2); patient right to be informed of AI use (¶3); the therapeutic decision is always reserved to the physician (¶5). Arts. 8–10 add research, data-processing and electronic-health-record provisions.

Editor rationale: Dedicated operative healthcare article with anti-discrimination, patient-information and human-decision rules.

AI-Driven Worker Displacement

Art. 12 establishes a national Observatory on the adoption of AI in the workplace charged with study, monitoring and technical support on the occupational, organisational and training effects of AI; Art. 11(1) frames AI as improving working conditions and productivity. Monitoring, not displacement protection.

Editor rationale: The law monitors workplace-AI effects via an Observatory but creates no substantive job-displacement protection.

International Coordination

Art. 1(2)/Art. 2 align the law with EU Reg. 2024/1689; Art. 19(3) requires the national strategy to take account of international humanitarian law; Art. 20(2) designates ACN as the single contact point with EU institutions under AI-Act Art. 70.

Editor rationale: EU-coordination and IHL references exist but the law creates no standalone international-cooperation mandate; coordination is mostly EU-implementation plumbing.

AI in Education

No operative schooling regime in force. Art. 24(2)(g) directs (as a delegation criterion) strengthening STEM/artistic competencies in school curricula; Art. 24(2)(i) requires AI-literacy training in universities/AFAM/ITS; Art. 15(4) promotes AI training for magistrates; Art. 22 supports youth.

Editor rationale: Education appears only as future delegation criteria and skills/AI-literacy promotion, not as operative present-law duties.

Synthetic Content Provenance

No standalone watermarking/provenance-marking duty in the law itself; provenance is reached only indirectly — Art. 612-quater criminalises deceptive AI-altered media (turning on whether content is apt to deceive as to genuineness) and the general transparency principle (Art. 4). Content-marking duties are left to the EU AI Act (Art. 1(2)).

Editor rationale: The Act addresses synthetic-media misuse but imposes no affirmative labelling/marking duty; provenance is implicit only in the deception element of Art. 612-quater.

Environmental Impact of AI Training

Art. 3(1) lists 'sostenibilità' (sustainability) among the binding general principles governing AI development and use, alongside transparency, proportionality, security and non-discrimination. No operative environmental-reporting or training-footprint duty.

Editor rationale: Sustainability is an enumerated principle but there is no operative environmental/energy provision targeting model training; borderline-silent, kept implicit/low.

Individual Redress

No general right to contest AI decisions. Art. 4(3) gives a right to object to authorised processing of one's personal data; Art. 16(3)(b) delegates the Government to provide compensatory/injunctive remedies and sanctions for training-data violations; the deepfake offence (Art. 612-quater) is prosecuted on the victim's complaint.

Editor rationale: Redress elements are partial (data-objection right, a delegated remedies regime, victim-complaint criminal route) and lack a general AI-decision contestation/appeal mechanism.

Sovereign AI Doctrine

No explicit sovereign-model/sovereign-compute mandate. Supported indirectly by Art. 5 (technological sovereignty + national-data-centre preference), Art. 19 (biennial national AI strategy, dual-use coordination with the Ministry of Defence) and Art. 23 (state investment in AI, cybersecurity and quantum computing).

Editor rationale: Capacity-building and localisation provisions support sovereign-AI ambitions but there is no operative sovereign-compute/sovereign-model rule, so implicit/low.