Skip to content
pitch
Deep Dives

The AI Act for businesses

The European AI Act now decides what a company may do with artificial intelligence, whether it builds a system, buys one or only lets staff use one. This deep dive explains the roles the law gives you, the four risk classes, what changes for people and for general-purpose models, and the dates that apply. It then turns to data protection, HR, contracts with vendors and a first governance set-up. It is written for readers who have heard of the AI Act and want to know what it means for the AI their organisation already uses.

Free with an account7 chapters18 lessons1 h 38 min

Course content

Free with an account. This overview is open to everyone. The lessons, media and downloads need a free account.

  1. Foundations

    3 lessons · 18 min

    What the law is for and who falls under it, from a start-up to a multinational outside Europe. You learn the four roles the law distinguishes and why the same company can hold more than one. The chapter ends with the moments when a buyer of AI is treated as its provider.

    1. What the AI Act is for and who it coversNew 6 min read
    2. Provider, deployer, importer, distributorNew 6 min read
    3. When you become the providerNew 6 min read
  2. Risk classes

    4 lessons · 22 min

    How a system is placed in one of four classes, from banned to nearly free of duties. You see which practices are prohibited, what makes a system high-risk and which systems only need to tell people that AI is involved. A short method for classifying your own tools ties it together.

    1. Classifying an AI systemNew 5 min read
    2. Prohibited practicesNew 6 min read
    3. High-risk AINew 5 min read
    4. Transparency duties and minimal riskNew 6 min read
  3. People and models

    2 lessons · 11 min

    The duty to make sure the people who work with AI understand it, and what that means in practice for a company of any size. Then the rules for general-purpose models, the large systems many products are built on. You see what their makers must do and what that means for the company that builds on top of them.

    1. AI literacyNew 5 min read
    2. General-purpose AI modelsNew 6 min read
  4. Duties of the user

    2 lessons · 11 min

    What a company owes when it uses a high-risk system that someone else built: oversight by people, logs, information for workers and for the people affected. A second lesson covers the assessment of effects on fundamental rights, who has to carry it out and how to go about it.

    1. The duties of deployersNew 6 min read
    2. The fundamental rights impact assessmentNew 5 min read
  5. Dates, supervision and data protection

    3 lessons · 17 min

    When each part of the law starts to apply, and what the Digital Omnibus changes to those dates. You see who supervises and fines in Belgium and what is still undecided there. The chapter closes with how the AI Act and the GDPR fit together when one system touches both.

    1. The timeline and the Digital OmnibusNew 6 min read
    2. Supervision, fines and BelgiumNew 6 min read
    3. The AI Act and the GDPRNew 5 min read
  6. HR and contracts

    2 lessons · 11 min

    Two areas where companies meet the law first. In recruitment and the management of staff, AI quickly falls in the high-risk class, and you see what that asks of an employer. You then look at the contract terms to ask of an AI vendor, and what to do when the vendor will not move.

    1. AI in HR and recruitmentNew 6 min read
    2. Contracts with AI vendorsNew 5 min read
  7. Governance

    2 lessons · 8 min

    A first set-up for managing AI in an organisation: an inventory, an owner, a few rules and a rhythm of review. A worked case then takes one company from the first question to a documented decision, so you can see the whole sequence on a single example.

    1. An AI governance starter kitNew 4 min read
    2. A worked case from start to finishNew 4 min read

Total time: 1 h 38 min

Read more in the Academy