Mandatory AI training in professional firms: what is already in force and what is not

Gaia Banfi
Formazione AI obbligatoria negli studi professionali cosa è già in vigore e cosa no
In this article
Want similar results?
Discover how LumenONE can transform your customer management.
Learn More
Back to Blog

On 10 June 2026 the Italian Council of Ministers approved the draft implementing decrees of the national law on artificial intelligence. The news fueled the idea that AI training only now becomes mandatory for professionals.

In reality the picture is more nuanced. There are three separate sources with three different dates, and distinguishing them helps clarify what is already binding today and what is still under approval.

The European obligation in force since 2025

AI training in professional firms did not begin in 2026. On 2 February 2025, Article 4 of EU Regulation 2024/1689, known as the AI Act, became applicable.

As a European regulation, it produces effects without the need for national transposition. The rule requires anyone who provides or uses AI systems to ensure a sufficient level of literacy among staff.

Translated into the daily life of a firm, the obligation concerns the organization as a whole. It involves the owner, collaborators and administrative staff, not only the professional registered with a board.

The ninety credits set by the national board

The second reference comes from the profession’s own rules. The regulation on continuing education adopted by the CNDCEC on 15 October 2025 has been in force since 1 January 2026.

The regulation requires registered members to earn ninety training credits over three years, at least nine of which in mandatory subjects. In Annex 1, under item B.1.2, appears the subject “Artificial intelligence in professional firms”.

As a result, hours spent training on AI count twice: as credits toward the board and as compliance with the European literacy obligation. For registered members this is already a current requirement.

What was actually approved on 10 June

The drafts approved on 10 June introduce, for regulated professions, AI training along three lines: technical, legal and ethical. Boards would have six months to update their training rules.

Within twelve months, the parameters of fair compensation should account for the risk classification of the AI system used. However, according to statement number 177 from Palazzo Chigi, this is a preliminary-stage approval.

The process still requires opinions from parliamentary committees, a return to the Council of Ministers, enactment and publication in the Official Gazette. Until then, these provisions are not in force.

The blind spot: the firm, not just the member

An often overlooked aspect concerns non-registered staff. According to the Work Trend Index, 78 percent of those who use AI at work bring their own tools, outside the organization’s control.

This phenomenon, called Shadow AI, carries significant weight in firms that handle client data. The internal training plan thus becomes the necessary complement to individual training credits.

Original article: accademiadeicommercialisti.it

Gaia BanfiLumenIA
I help Italian companies understand and adopt artificial intelligence in a concrete, safe, and measurable way.

You might be interested

See all
    L intelligenza artificiale nella ristorazione come i ristoranti affrontano costi e nuovi consumi
    • News

    Artificial Intelligence in Restaurants: How the Food Sector Faces Rising Costs and New Consumption

    The out-of-home dining sector is going through a deep transformation. Customers visit the various types of venues less often, while off-premises sales channels and large-scale retail are gaining market share.…

    ⏱ 3 minuti di lettura
    AI e lavoro le imprese italiane cercano 621mila esperti, oltre la metà introvabili
    • News

    AI and Jobs: Italian Companies Seek 621,000 Experts, More Than Half Are Unfindable

    Italian companies are accelerating their investments in artificial intelligence, but they struggle to find the professionals needed to manage it. This is the alarm raised by Confartigianato in an analysis…

    ⏱ 2 minuti di lettura
    GPT 5.6 Sol, Terra e Luna OpenAI presenta in anteprima la nuova serie di modelli
    • News

    GPT-5.6 Sol, Terra and Luna: OpenAI Opens a Limited Preview of Its New Model Series

    OpenAI has begun a limited preview of the GPT-5.6 series, made up of three distinct models. Sol is the flagship model, Terra is aimed at everyday work, and Luna focuses…

    ⏱ 3 minuti di lettura