The EU AI Act isn't a future event — key parts are already enforceable, and the penalties are serious. If you build or ship AI that touches EU users, the implementation timeline matters now.
The dates that are already live
- Feb 2, 2025 — bans on prohibited AI practices and AI-literacy obligations took effect.
- Aug 2, 2025 — rules for general-purpose AI (GPAI) models and the penalty regime kicked in: fines up to €35M or 7% of global turnover for prohibited practices, €15M / 3% for other breaches.
What's coming
- Aug 2, 2026 — full application for high-risk systems: conformity assessments, technical documentation, CE marking, EU database registration.
- Aug 2, 2027 — pre-2025 GPAI models must be compliant; extended transition (to 2028) for AI embedded in regulated products.
The Act is risk-tiered, not blanket. The first job isn't compliance — it's classification: which of your systems are prohibited, high-risk, or limited-risk?
For most teams the practical work is data governance, human oversight, transparency and documentation — the same disciplines that make AI trustworthy regardless of jurisdiction.
Sources
- EU Artificial Intelligence Act — Implementation Timeline
Written by ivector
Start a project →