Regulation 101

Understanding the EU AI Act

The world's first comprehensive AI law. It doesn't ban AI — it sorts AI by how much harm it could cause, and sets stricter rules the higher the risk. Here's what you need to know in ten minutes.

01

Who does it apply to?

Any organisation that builds, sells or uses AI in the EU — even if it's based elsewhere. Your obligations depend on your role.

Provider

Develops an AI system or has it developed, and places it on the market under its own name.

A fintech building and selling a credit-scoring model to banks.

Deployer

Uses an AI system under its own authority in a professional context.

A bank using that credit-scoring model to approve loans.

Importer & Distributor

Brings non-EU AI systems into the EU market or makes them available.

A reseller offering a US fraud tool to Nordic insurers.

02

Four risk levels, one principle

The higher the risk to people's safety and rights, the stricter the rules. Unacceptable risk is banned, high risk is heavily regulated, limited risk needs transparency, and minimal risk is free to use.

Explore the four risk levels →

03

Timeline

  1. Aug 2024

    The AI Act enters into force.

  2. Feb 2025

    Prohibited practices banned. AI literacy duty for staff begins.

  3. Aug 2025

    Rules for general-purpose AI models (e.g. large language models) apply.

  4. Aug 2026

    Most obligations apply, including high-risk systems listed in Annex III such as credit scoring.

  5. Aug 2027

    High-risk rules for AI embedded in regulated products apply.

04

What high-risk systems must do

In finance, credit scoring and life/health insurance pricing are named high-risk. These are the core requirements:

1

Risk management

A continuous process to identify, assess and reduce risks across the system's lifecycle.

2

Data governance

Training data must be relevant, representative and checked for bias.

3

Technical documentation

Detailed records proving the system meets the requirements.

4

Logging

Automatic event logs so decisions can be traced afterwards.

5

Transparency to deployers

Clear instructions on capabilities, limits and correct use.

6

Human oversight

People must be able to understand, override or stop the system.

7

Accuracy & robustness

Consistent performance and protection against errors and attacks.

8

Conformity & registration

Assessment before launch, CE marking and registration in the EU database.

05

Obligations for everyone

AI literacy

Staff who use AI must understand it well enough to use it responsibly — e.g. advisers knowing a model's limits.

Transparency

People must know when they talk to a chatbot, and AI-generated content must be labelled.

06

Penalties

Fines are the higher of a fixed amount or a share of global annual turnover.

€35M or 7%

Prohibited practices

Using banned AI, such as social scoring.

€15M or 3%

Other obligations

Failing high-risk or transparency requirements.

€7.5M or 1%

Incorrect information

Supplying misleading information to authorities.

Where does your AI system fit?

Answer ten quick questions to see its likely risk level.

Take the self-assessment