The Artificial Intelligence Act: Taking Normative Imbalances Seriously
The final text for the Artificial Intelligence Act (AIA) was approved by the European Council on May 21, 2024. This pivotal regulation is set to become generally applicable from August 2, 2026. The AIA aims to enhance the functioning of the internal market and promote the adoption of human-centric and trustworthy artificial intelligence (AI), while ensuring protection for health, safety, and fundamental rights. It seeks to support innovation without compromising on essential values.
Introduction
The AIA establishes a framework for AI systems to ensure that fundamental rights are upheld while fostering economic growth. This legislation is crucial in addressing the potential risks associated with AI, recognizing that inadequate regulation could undermine significant societal values.
Rule-based vs. Principle-based Regulations
Legal regulations must strike a balance between normative means and the goals they aim to achieve. This balance can be categorized into rule-based and principle-based regulations:
- Rule-based regulations: These are generally precise and predictable but may lack adaptability to changing circumstances. They consist of rigidly formulated rules that can lead to suboptimal outcomes if not adequately defined.
- Principle-based regulations: These regulations are more abstract and flexible, allowing for a balancing act between competing values. However, their application can be contentious and vary across contexts.
The Normative Structure of the AIA
The AIA predominantly employs a rule-based structure, classifying AI systems based on their associated risks. The legislation includes:
- Unacceptable risks: Certain AI applications are prohibited outright.
- High-risk systems: These systems are subject to stringent requirements aimed at protecting fundamental rights.
- Moderate and minor risks: These categories face less rigorous regulations.
For high-risk systems, the AIA mandates the establishment of a risk management system, incorporating elements that require contextual interpretation, such as “reasonably foreseeable risk.” Non-compliance can result in significant administrative fines.
Imbalanced Regulation
Despite its comprehensive approach, the AIA faces challenges that may hinder its effectiveness:
- The measures to safeguard fundamental rights may be insufficient, especially given the rapid evolution of AI and limited case law.
- Governance structures may impose excessive burdens on obligated parties, potentially leading to overcompliance.
- AI providers might exploit vague regulations to avoid high-risk classifications, undermining protections for fundamental rights.
Concluding Remarks
Addressing the identified challenges within the AIA is crucial for achieving its goals. Potential measures include:
- Expanding AI regulatory sandboxes to assess impacts on fundamental rights in controlled scenarios.
- Encouraging alternative dispute resolution methods to alleviate stringent demands on high-risk systems.
- Implementing expert court panels to expedite proceedings related to high-risk systems.
In conclusion, while the AIA sets a foundational framework for AI regulation, it requires ongoing refinement to ensure that it meets its ambitious objectives without stifling innovation.
