Executive Summary

Gabriele Mazzini, the chief architect of the EU's artificial intelligence law (AI Act), has resigned in protest against the final version and accuses the EU of creating an unusable regulatory framework under panic and time pressure. The originally streamlined law (85 articles) was inflated into a complex monster (113 articles, 180 preambles) that hinders innovation rather than promoting it. Mazzini's resignation reveals fundamental weaknesses in the EU legislative process and shows how technological alarmism leads to regulatory overreaction.

Critical Key Questions

Regulatory Panic: How can democratic institutions prevent apocalyptic AI narratives from leading to hasty laws that harm innovation more than protect it?
Expert Authority vs. Political Time Pressure: Why did the EU Commission ignore its own chief expert and sacrifice professional competence for political prestige thinking?
Competition Distortion: Don't unclear laws create exactly the legal uncertainty that large tech corporations can exploit while deterring smaller innovators?

Scenario Analysis: Future Perspectives

Short-term (1 year):
AI companies in the EU will invest massively in compliance teams. Legal uncertainty leads to investment freezes at smaller firms, while large corporations expand their legal departments.

Medium-term (5 years):
Brain drain of AI talent to less regulated markets (USA, Asia). The EU falls further behind technologically, while other regions benefit from more pragmatic regulatory approaches.

Long-term (10-20 years):
Europe becomes the "Museum of Digitalization" – technologically dependent, but regulatorily leading in irrelevant standards. The intended "Brussels Effect" fizzles out due to lack of innovative EU companies.

Main Summary

Core Theme & Context

Gabriele Mazzini, main author of the EU AI Act, has quit his well-paid Commission position because he considers the final law "failed." After the ChatGPT hype in 2023, his originally lean, risk-based regulatory approach was transformed under political time pressure into a complex, impractical regulatory framework.

Most Important Facts & Figures

Law inflation: From 85 to 113 articles, from 89 to 180 preambles
Development time: Only 3 years (vs. 20 years for GDPR)
Resignation timing: August 1, 2024, exactly on the day of entry into force
Original idea: Risk-based AI regulation by product categories
Turning point: ChatGPT release led to "apocalyptic scenarios" and regulatory panic
Mazzini's protest: Already warned internal supervisors in vain in fall 2023

Stakeholders & Affected Parties

Losers: EU AI startups, innovators, medium-sized technology companies
Beneficiaries: Large tech corporations with strong legal departments, consulting firms
Affected institutions: EU Commission (reputational damage), European AI research, venture capitalists

Opportunities & Risks

Risks:

Legal uncertainty due to vague formulations
Innovation flight from the EU to less regulated markets
Competitive disadvantage against USA and China
Bureaucratic paralysis through hundreds of additional directives

Opportunities:

Course correction possible: Mazzini calls for halt and substantial revision
Learning effect for future tech regulation
Focus on specific legal gaps instead of mammoth laws

Action Relevance

Immediate measures: Companies must develop compliance strategies, but without legal certainty. Political decision-makers should take Mazzini's reform proposals seriously.

Strategic positioning: The EU must choose between "regulation as location advantage" and "over-regulation as innovation brake." A course change is still possible, but requires political courage.

Sources

Primary Source:
Main Author of EU AI Act Speaks Out – NZZ

Verification Status: ✅ Facts checked on 12.09.2025

⚠️ Transparency Note: This analysis is based on Mazzini's subjective perspective. Official statements from the EU Commission regarding his accusations are not available.