This is an EU Artificial Intelligence Act summary to help you get an overview of the EU’s comprehensive AI governance framework. It is also the world’s first law on AI specifically. This is a summary of:

  • why it’s important;
  • who it applies to;
  • when it takes effect and,
  • what actions you should take.

This article also provides you with useful links so you can read more about the AI Act.

EU Artificial Intelligence Act summary in plain language | Infographic

For a visual summary, we’ve created a plain language guide on the AI Act infographic. It is to help you visualise and understand the key aspects of the Act. We will publish it soon.

Why do we need the EU Artificial Intelligence Act?

As AI continues to rise and reach new heights, regulatory authorities must actively manage the risks associated with the development and deployment of AI. The EU AI Act is vital for the ethical and responsible use of AI. It ensures transparency, accountability, and fairness in AI systems. This is to address concerns about bias and promoting safety standards. With clear legal guidelines, it aims to harmonise AI regulations across the EU, fostering international competitiveness and positioning the EU as a global leader in responsible AI development.

Who does the EU AI Act apply to?

  • Organisations that use AI systems in their engagement with people living in the EU because the AI Act applies extraterritorially, meaning it applies to people outside of the EU as well as those based in the EU.
  • Developers of AI and providers of AI because the AI Act creates additional transparency, disclosure and record-keeping requirements.
  • Users of AI because it will impact the type of AI you can interact with, how AI is made available and what safeguards you should expect from AI providers.

How does the EU Artificial Intelligence Act work?

The EU AI Act takes a risk-based approach by classifying all AI systems into one of four categories.

  1. Unacceptable risk
  2. High risk
  3. Limited risk
  4. Minimal risk

AI systems are categorised by considering their link to the values of the EU, fundamental human rights and the well-being of natural persons. Thus, the riskier the AI system is deemed to be based on its interaction with these elements, the more regulated it is.

When does it take effect?

The EU AI Act takes effect 20 days after the Official Journal of the EU published it.

EU AI Act timeline

There will be varying compliance grace periods depending on the type of AI system.

  • 6 months after the AI Act enters into force unacceptable risk AI will be prohibited.
  • 12 months after the AI Act enters into force the providers of general-purpose AI will incur obligations.
  • 24 months after the AI Act enters into force all other obligations will apply.

Actions you can take