Data, AI & Digital Sovereignty
10. Bruegel calls for EU AI Act recalibration toward liability, incident reporting and ex-post supervision
Bruegel argues that the EU AI Act should move from a predominantly ex-ante product-safety model toward a hybrid model that lowers up-front burden for many suppliers while adding stronger AI liability, post-deployment monitoring, universal transparency, researcher access, near-miss reporting and a public incident registry.
BY GEOPOLITICS DESK · JUNE 11, 2026 · 1 MIN READ
Bruegel argues that the EU AI Act should move from a predominantly ex-ante product-safety model toward a hybrid model that lowers up-front burden for many suppliers while adding stronger AI liability, post-deployment monitoring, universal transparency, researcher access, near-miss reporting and a public incident registry. It recommends reviving an AI liability framework and using tiered requirements based on deployment scale and harm profile.
The legal advisory signal is clear: AI compliance will not stop at conformity assessments. Liability allocation, audit access, incident taxonomy, trade-secret protection and regulator-facing evidence trails will become part of AI contracting and governance.