EU AI Act
The EU's risk-based AI law
The European Union’s risk-based AI law. Prohibitions and AI-literacy duties began applying in February 2025, general-purpose-model obligations in August 2025, and most other provisions in stages. Under the current implementation timetable, certain Annex III high-risk uses apply no later than 2 December 2027, while systems embedded in regulated products apply no later than 2 August 2028.
Concrete example
A provider first determines which part of the Act applies; obligations differ for prohibited uses, high-risk systems, transparency duties, and general-purpose models.
Why it matters
A major compliance framework whose dates and guidance can change; check the Commission’s current timeline before acting.
Nearby in Safety & Alignment
Evidence
Sources for this definition
- 1AI Act application timeline
European Commission · Regulator · checked 2026-07-13