Jump to content

Italian AI Act implementation decree (2026)

From AI Law Wiki

AI-generated text. This page was generated using artificial intelligence.

Law & policy information
Name
Italian AI Act implementation decree.[1]
Jurisdiction
Italy.[1]
Instrument type
Legislative decree.[1]
Bill / legal citation
Legislative Decree No. 179 of October 7, 2026; Gazzetta Ufficiale, General Series No. 234, October 8, 2026.[1]
Legislative or adoption status
Issued October 7, 2026; published October 8, 2026.[1]
Effective dates
Entered into force October 23, 2026.[1]
Last source verification
October 9, 2026.

Italian AI Act implementation decree (2026) is Legislative Decree No. 179 of October 7, 2026. Published in the Gazzetta Ufficiale on October 8, it entered into force on October 23, 2026. It adapts national law to the EU AI Act, addressing regulators’ powers, enforcement and AI training.[1]

The Italian-language original, including all 54 articles, was reviewed October 9, 2026.

Authorities and enforcement

AgID is the notifying authority; ACN is the general market-surveillance authority and single point of contact. Sector allocations cover the Bank of Italy, CONSOB and IVASS, while preserving the data-protection authority’s competence for specified uses.[2][3][4]

The decree provides inspection, corrective and administrative-penalty procedures. Article 23 implements the AI Act’s tiered penalties, including up to €35 million or 7% of worldwide annual turnover for prohibited practices, subject to the applicable rules for smaller enterprises. It also provides national penalties for breaches of fundamental-rights impact-assessment and explanation duties.[5]

Sandboxes and training

Articles 26–31 establish the national regulatory-sandbox framework under AgID and ACN, including admission, supervision, cross-border cooperation and reporting. Implementation measures govern its operation.[6][7][8][9]

Education, public administration, justice and professional-training provisions address AI literacy, technical limitations, oversight and legal responsibilities. Professional bodies must adapt their training arrangements within six months of commencement.[10][11][12][13]

Worker protections

Article 41 reserves final employment decisions to a natural person exercising effective, autonomous authority. It covers creation, alteration and termination of employment, disciplinary measures and performance evaluation; candidate search and selection are excluded from its definition of final hiring decisions. Workers can request an intelligible explanation of AI’s contribution and principal parameters. Decisions violating the human-decision requirement are void.[14]

Article 42 requires relevant workplace AI risks to be addressed in occupational risk assessment, with information and training for workers.[15]

Trade secrets

Article 52 expressly includes qualifying AI training data, algorithms and mathematical methods within the Industrial Property Code’s protected business information, where the existing statutory conditions are met.[16]

References