Archive · 2 Aug 2026
EU AI Act Goes Live, Targets Chatbots, Deepfakes, High-Risk Systems
The European Union has made a new law called the AI Act that is now in force.
It says that if you make or use a computer program that can talk to people or create fake pictures, you must tell people it is a robot.
The law also says some AI programs are too dangerous and are banned.
Big AI programs that can do many things must be tested and have rules about how they are made.
The EU will check that companies follow the rules and can fine them if they don’t.
This helps keep people safe and makes sure AI is used in a good way.
EU AI Act now in force, setting rules for chatbots, deepfakes, and high‑risk AI.
Developers must label AI interactions and deepfakes with visible tags and watermarks.
AI systems are classified into four risk tiers, from unacceptable (banned) to minimal (unregulated).
General‑purpose AI models face strict documentation, testing, and potential fines up to a percentage of global turnover.
Enforcement will be carried out by the AI Office, national authorities, and the European Data Protection Supervisor, with new complaint tools.
- Who
- European Union Commission and its AI Office
- What
- The EU AI Act goes live, establishing rules for AI systems
- When
- Effective immediately, with phased implementation through 2028
- Where
- European Union
- Why
- To ensure safe, transparent, and ethical AI and protect public trust
- Effective date
- 2026-08-02
- Risk tiers
- 4
- Enforcement bodies
- AI Office, National Competent Authorities, European Data Protection Supervisor
- Fine cap
- Percentage of global turnover
- Targeted AI types
- Chatbots, deepfakes, high-risk AI