Brussels Effect or Experimentalism? The EU AI Act and Global Standard-Setting
Abstract
Supporters of the EU’s Artificial Intelligence Act have presented it as a potential global standard for AI regulation, reminiscent of the widely recognized ‘Brussels effect’. This paper contrasts this expectation with the alternative of experimentalist governance, which views the EU’s AI Act as one of several regulatory approaches and advocates for a more collaborative interaction with global regulatory frameworks. The analysis unfolds along two main lines: the inherent uncertainties of AI compared to established digital technologies and the procedural nature of the AI Act itself, suggesting that its external impact aligns more with experimentalist governance than the Brussels effect.
Theorizing EU Digital Regulation
The EU’s regulatory role in a globalized context can oscillate between being a rule-setter or a rule-taker. The EU is well-positioned to set global standards due to its significant market size and established regulatory capacity. The Brussels effect denotes the EU’s ability to influence global standards through high regulatory demands, compelling other jurisdictions to adopt similar regulations to access the EU market. This phenomenon has been observed with the General Data Protection Regulation (GDPR), which has inspired similar laws worldwide.
The AI Act: An Experimentalist Approach
The EU AI Act, adopted in 2024, is heralded as a significant step in AI regulation, with the EU positioning itself as a global frontrunner. However, this article argues that the AI Act’s procedural framework and the uncertainties surrounding AI technology may limit its potential to create a straightforward Brussels effect. The Act adopts a risk-based approach, where regulations are more stringent for higher-risk applications, yet it remains procedural and open-ended, promoting a form of experimentalist governance instead.
Differences in AI Regulatory Challenges
AI presents unique regulatory challenges compared to established digital technologies. Key differences include:
- Fundamental Uncertainty: AI raises a wide range of uncertainties, requiring regulators to take an interventionist stance to balance its promises and perils.
- Trust in Regulation: Regulation is not merely a burden for companies; it serves as a critical tool for instilling consumer trust in AI products.
- Market Structure: The AI market is expected to be more fragmented compared to the concentrated structure of the internet economy, making the Brussels effect less likely to manifest.
Contents of the AI Act
The AI Act is characterized by its procedural nature, allowing for flexibility and adaptation in response to evolving AI challenges. It features:
- High-Risk AI Systems: A significant focus is placed on high-risk applications, which require a comprehensive risk management system.
- General-Purpose AI Models: A distinct regulatory strategy is applied, with a central role for the newly created AI Office within the European Commission.
- Transparency Obligations: The Act mandates transparency measures, such as bot disclosure, to inform users about AI-generated content.
Conclusion
The EU AI Act represents an ambitious attempt to regulate AI, but its procedural and experimentalist orientation suggests that it may not achieve the definitive global standard implied by the Brussels effect. Instead, the Act embodies a learning process where the EU is engaged in a collaborative effort with other jurisdictions. The future of AI regulation is likely to be characterized by ongoing dialogue and shared standards rather than strict competition among regulatory powers.
