The European Parliament and the EU's 27 member states will need to approve the AI Act, which seeks fines of up to €35M or 7% of global turnover for violations
- Region couldn't ‘let the perfect be the enemy of the good’ — Negotiators broke coffee machine working late-night deal
Context & Ripple Effects
The late-stage deal follows the Parliament’s earlier push for copyright-use disclosure requirements for chatbot makers, indicating that transparency obligations were part of the Act’s contested scope, not an add-on.
This agreement moves the file to formal approval by Parliament and member states. Subsequent coverage records both a member-state agreement paired with an AI Office and Parliament’s eventual approval of the risk-category framework, placing this moment at the transition from negotiation to institutional adoption.
First-order effects
- Parliament and the 27 member states become the immediate gatekeepers: the political deal cannot become the AI Act until they formally approve it.
- The proposed penalty ceiling—up to €35 million or 7% of global turnover—raises the immediate compliance stakes for organizations that could be found in violation once the regime is in force.
Second-order effects
- AI developers and deployers face pressure to map their systems and documentation to the Act’s eventual obligations before enforcement, particularly where transparency requirements are implicated.
- The prospect of turnover-based fines makes implementation and enforcement design commercially consequential, pushing firms to treat EU AI governance as a board-level legal and operational issue rather than a narrow product-policy matter.
Third-order effects
- If the risk-category approach holds, AI regulation shifts toward a durable model in which obligations and enforcement exposure vary by a system’s use and risk profile.
- The negotiation also illustrates the continuing tension between enforceable safeguards and competitiveness concerns; later attempts to seek exemptions for regulated sectors suggest that scope and enforcement will remain contested even after headline legislation advances.
The trend: AI governance is moving from voluntary principles toward enforceable, risk-based regimes with penalties large enough to shape how AI products are designed and deployed.