On 23 October 2026, Italy’s Legislative Decree No. 179 came into force. It implements the European Union’s Artificial Intelligence Act at national level, and much of it concerns not technology but people: who must be trained to use AI, by whom, and with what content.
The decree arrives three months after the EU loosened its own AI literacy rule. That contrast makes Italy a useful test of a practical question facing institutions everywhere: whether AI competence should be left to employers’ discretion or built into the formal systems that train and license professionals.
What the decree is
Legislative Decree No. 179 of 7 October 2026 was published in the Gazzetta Ufficiale (Italy’s official journal) No. 234 on 8 October 2026. Its title describes it as adapting national law to Regulation (EU) 2024/1689, the AI Act. According to Quotidiano Sanità, it has 54 articles and also implements Italy’s 2025 national AI law, Law 132/2025.
The decree sets up enforcement. The Agency for Digital Italy (AgID) becomes the national notifying authority and the National Cybersecurity Agency (ACN) the market surveillance authority (Articles 4 and 5). Article 23 sets fines of up to €35 million or 7% of worldwide annual turnover for prohibited AI practices, and up to €15 million or 3% for breaches of other listed obligations.
Analysts describe a “multi-authority” model. Writing in Agenda Digitale on 15 October 2026, data-protection lawyer Nadia Martini notes that the Bank of Italy, the financial markets regulator Consob, the insurance regulator IVASS and the data protection authority each keep roles in their own sectors.
Training as a legal duty
The provisions that stand out most are those on education and training. Rather than a single literacy requirement, the decree assigns responsibilities to specific institutions.
Professions
Article 47 requires professional bodies (the ordini professionali that regulate lawyers, engineers, doctors and others) to include AI literacy and training in both initial and continuing education. The courses must cover technical and legal content and the ethical responsibilities of professionals who use AI systems. Each must meet a minimum number of hours per continuing-education period, and the bodies have six months to adapt their rules.
Health professionals
Health is singled out. As reported by Quotidiano Sanità, Article 51 reserves a share of each health professional’s three-year continuing medical education (ECM) obligation for AI competences, covering how systems work, professional responsibility and duties to inform patients. The number of credits has not yet been set. Agenas, the national agency for regional health services, must also identify AI training for health managers. Both are due by 23 April 2027.
Universities, schools and public servants
Under Article 44, universities must promote training on how AI systems work, including their limits and risks, critical interpretation of outputs and cybersecurity. Courses with a mainly scientific, technological, medical or biomedical focus must integrate ethical and legal content.
For schools, the decree provides for teacher training and student programmes. Quotidiano Sanità reports up to €100 million for a plan on the risks of unaware use of digital platforms, including AI-based ones, under the 2021-2027 national schools programme.
Article 40 asks the Minister for Public Administration to set a minimum AI competence framework for public employees, delivered with the National School of Administration and Formez PA. Article 46 requires training for justice staff to include the skills needed for human oversight where high-risk systems are used.
Employers
Two articles apply to the workplace. Article 41 states that a final decision on hiring, changing or ending employment, discipline or performance evaluation is “in every case reserved to a natural person” with real and independent authority. Quotidiano Sanità notes that recruitment and candidate selection are excluded from this rule. Article 42 requires employers to include AI systems that affect work organisation or safety decisions in their statutory risk assessment, and to inform and train workers on the specific risks.
A different direction from Brussels
Italy’s approach contrasts with recent EU changes. The AI Act’s original Article 4, which applied from 2 February 2025, required providers and deployers of AI to “ensure” a sufficient level of AI literacy among their staff.
Regulation (EU) 2026/1744 of 8 July 2026, part of the EU’s “digital omnibus” simplification package, rewrote that article. According to an analysis of the Official Journal text by NicFab, organisations must now “take measures to support the development” of AI literacy, and the text states that this “does not require providers or deployers to guarantee any specific level” of literacy. The change applied from 27 July 2026. The same regulation postponed most high-risk obligations to 2 December 2027 and 2 August 2028.
The omnibus was presented as a simplification package, and the revised article directs the Commission and member states to support operators, especially small and medium-sized enterprises. Italy has instead used its own education and licensing systems to require AI training for whole professions and sectors.
What remains uncertain
Much of the decree depends on rules not yet written. Minimum hours for professional training, the share of ECM credits and the ministerial decree on school training are all still to come. Agenda Digitale notes that the 2025 ministerial guidelines for schools continue to apply in the meantime, and that the regulatory sandbox exists in law but is not yet operational.
There is also an open question about whether mandated hours produce real competence. Recent evidence suggests that design matters. A randomised trial of Oak National Academy’s Aila lesson-planning assistant, published by the Education Endowment Foundation in October 2026, involved 416 teachers in 108 primary schools in England. It found that teachers using the tool saved about 49 minutes a week on preparation with no loss of resource quality, but used it for only about 7% of lessons. The foundation stressed that teachers still need to check AI outputs and raised concerns about “cognitive offload” and its effect on professional expertise.
That finding speaks to what training must cover. Knowing how to operate a tool is the easy part. Knowing when to trust it, how to check it and who is answerable for the result is harder, and is what the decree’s emphasis on ethics, responsibility and human oversight attempts to address.
Why it matters beyond Italy
Italy’s decree applies only in Italy, but the questions it raises are general. Health systems, universities and public agencies in many countries are adopting AI tools and must decide how their staff will learn to supervise them. Italy has chosen to make that training part of how professionals qualify and stay qualified, rather than leaving it to each employer.
Whether the approach works will depend on implementation: the content of courses, how learning is assessed and whether training reaches the people who actually use the systems. These are questions of practical, applied professional learning and of accountable, human-centred AI, the same principles set out in how Praxis uses AI and explored in the earlier insight You cannot sue an algorithm.
Praxis Global Institute designs applied training, including the International Professional Readiness Program, that helps professionals and institutions build the judgement to use AI responsibly; contact the team to discuss how this applies to your organisation.
Praxis Editorial Team
The Praxis Editorial Team publishes Praxis Global Institute’s explainers, practical guides and analysis of new evidence and policy in global health, education and professional capability, ethical AI and technology, research and data, and institutional strengthening. Every article is researched from primary sources, cites its evidence and is published under the editorial direction of its editors.
Facing a similar challenge in your institution?
We work with teams on research, data, digital systems, and learning that hold up in practice.
