Learn AI · Level 4 · Expert view · step 5 of 13 · word 2 of 2
What is the EU AI Act?
The EU AI Act is the European Union's law on artificial intelligence, the first broad AI law of its kind. It sorts AI by risk, bans some uses outright and sets rules for the rest.
In plain English
The EU AI Act regulates AI according to how much harm it could do, rather than by technology. Think of four levels:
- Unacceptable risk: banned. For example, “social scoring” of citizens, AI that manipulates people in harmful ways, and most real-time facial recognition by police in public spaces.
- High risk: allowed with strict rules. AI used in hiring, credit scoring, education, medical devices or law enforcement must be tested, documented, overseen by people and checked for bias.
- Limited risk: be transparent. People must be told when they are talking to a chatbot, and AI-generated content such as deepfakes must be labelled.
- Minimal risk: no new rules. Spam filters, video-game AI and most everyday tools.
Makers of large general-purpose models, such as the ones behind ChatGPT or Gemini, have their own duties, including publishing a summary of their training data and, for the most powerful models, testing for serious risks.
Why it matters
The law applies to any company whose AI is used in the EU, wherever the company is based, so it shapes products worldwide, much as the EU’s privacy law (GDPR) did. Fines for the worst breaches can reach €35 million or 7% of a company’s worldwide annual turnover.
It is coming in gradually. It became law in August 2024; the bans have applied since February 2025, and the rules for general-purpose models since August 2025. In 2026 the EU agreed to delay the strictest high-risk rules: to December 2027 for systems such as hiring and credit scoring, and to August 2028 for AI built into regulated products such as medical devices and toys.
An example
A bank uses AI to decide who gets a loan. Under the Act this is high risk, so the bank must show the system was tested for unfair bias, keep records of how it works, let a person review decisions and tell customers that AI was involved.
What to watch out for
Dates and details are still moving, with guidance and standards being written as the rules start to apply, so check official EU sources before relying on a specific deadline. Other places are taking different approaches: the US has no single federal AI law, China has rules for specific uses such as generative AI and deepfakes, and the UK relies largely on existing regulators.