Welcome. Generative artificial intelligence has entered organisations faster than any previous technology, often before anyone decided to let it in. This learning path is here to put things back in order: to understand what these tools really are, to learn to work with them methodically, to know the rules Europe has written and to know which risks they bring with them.

The thread is twofold. On the method side, the path takes up and systematises the approach of the three Academies of the main providers: Anthropic's AI Fluency framework with its four competencies, the OpenAI Academy path from the fundamentals to working with agents, and Google's prompting method. On the rules side, the reference is Regulation (EU) 2024/1689 (AI Act), read together with Italian Law 132/2025 and Regulation (EU) 2026/1744, the Digital Omnibus, in force since 27 July 2026, which rewrote its timetable and amended some obligations on the merits.

Learning path objectives

By the end of the learning path you will be able to:

• tell artificial intelligence, machine learning, deep learning and generative AI apart, and recognise what falls within the legal definition of an AI system (Module 1);
• explain how a language model produces an answer, why it can be wrong with confidence and what its structural limits are (Module 2);
• apply a working method with AI: when to delegate, how to describe the task, how to assess the result and how to remain responsible for what you deliver (Module 3);
• find your way among assistants, AI in office tools, retrieval from company documents and agents, and choose the right tool for the problem (Module 4);
• place your organisation within the AI Act: what role it holds, what risk level the systems it uses fall into, which practices are prohibited (Module 5);
• recognise the obligations for high-risk systems and for general-purpose models, apply the transparency of Art. 50, read the application timetable correctly and make informed use of the guided questionnaire with which a system is placed within the regulation (Module 6);
• manage the security and data protection risks introduced by AI and contribute to a usable company policy (Module 7).

Who it is for, and prerequisites

No technical or programming skills are needed: the learning path explains the concepts from scratch. Having followed the learning paths on security fundamentals and on compliance beforehand is useful, but not compulsory: modules 5, 6 and 7 take the concepts of risk, control and personal data processing as given, although they are recalled wherever needed.

How it is run

The path is delivered entirely online, asynchronously: you can study when you want, at your own pace, from any device. The estimated total duration is 31 hours, which you can spread out freely over time.

Each module is organised into four lessons and two tests:

1. Lesson. A reasoned treatment of the topic, with precise references, diagrams and concrete cases.
2. In depth. The updated legal and technical picture, with the deadlines that matter and references that can be checked at source.
3. Real-world cases. What actually happens on the market: documented episodes always read with the same grid, down to the control that would have broken the chain.
4. In practice. What you do on Monday morning: procedures, checklists, ready-made templates and indicators to measure the result.
5. Test. Closed-answer questions on the whole module. Passing requires at least 70%; attempts are unlimited and the best mark counts. At the end of every attempt you receive a detailed explanation of each answer, including the correct ones.
6. Exercise. A case to work through by deciding: every choice opens a different path and the outcome depends on what you chose, with an explanation of what would have happened otherwise. Marking is automatic and you can retake it as many times as you like.

Progression. The 7 modules are taken in sequence: each module unlocks only after you have passed the previous module test. The exercises do not block progress, but they are an integral part of the path and of the overall assessment.

Final test and certificate

Once the tests of all the modules have been passed, the final test is unlocked: 30 questions drawn at random from the topics of the whole learning path, with the same 70% pass mark and unlimited attempts. On passing it you obtain the learning path certificate, downloadable as a PDF, with a unique code that allows its authenticity to be verified.

Warning

The contents are up to date as at August 2026. Two warnings specific to this subject. The first: the regulatory framework is moving and the Digital Omnibus has moved some AI Act deadlines beyond their original date, so before planning a compliance step always check the consolidated text on EUR-Lex and the Commission's communications: this also applies to the public self-assessment tools, which stay fixed at the version of the text they were built on, as Module 6 shows in concrete terms. The second: AI tools change names, prices and capabilities within weeks, so the path teaches the method and the selection criteria, not the list of this month's products.

Independence from providers

The path cites the Academies of Anthropic, OpenAI and Google as public and free teaching sources, and takes up their methods because they are the reference training material available today. NormaShield Academy is not affiliated with any of these providers, does not resell their products and does not recommend one tool over another: the selection criteria you will find in Module 4 are designed to be applied to any tool, including those that do not yet exist.

Support

In the Announcements forum you will find messages from the teacher. For questions about the content or the exercises use the channels indicated by your training contact.