The world's first comprehensive law for artificial intelligence. It sorts AI systems by risk - banned, high-risk, limited - and sets hard obligations for each tier.
It binds anyone who builds or deploys AI touching the EU, the duties phase in on fixed dates, and fines reach 7% of global turnover. It is law, not guidance.
If your organisation uses or sells AI in Europe, this is a compliance clock on the board's agenda - and the real question is whether the EU can enforce what it wrote.
This instrument tracks how far the EU has actually stood the law up - designations, deadlines, fines - versus what it announced. Updated when the regulatory record changes.
The delay is law. Regulation (EU) 2026/1744 was published in the Official Journal on 24 July and entered into force on 27 July, six days before the deadline it was racing. Article 50 transparency was not moved and applies from 2 August, alongside the Commission's first power to fine a general-purpose model provider. Fines issued stayed at zero.
| Obligation | Article | Status | Effective |
|---|---|---|---|
| Prohibited practices | Art 5 | in force | Feb 2025 |
| General-purpose AI rules | in force | Aug 2025 | |
| Transparency and disclosure | Art 50 · marking grace to Dec 2026 | upcoming | 2 Aug 2026 |
| Commission fines for GPAI providers | Art 101 | upcoming | 2 Aug 2026 |
| High-risk, Annex III | use-based | moved | Dec 2027 |
| High-risk, Annex I | product-embedded | moved | Aug 2028 |
We take positions only where the evidence supports them - and grade them in public.
How we build research - the method →