In a historic move, the European Union Parliament has enacted the Artificial Intelligence Act, the world’s first comprehensive legal framework designed to govern the development and deployment of artificial intelligence (AI) systems. The EU Parliament approved the act on March 13, 2024, after nearly three years of debate and deliberation. This groundbreaking legislation establishes risk-based classifications for AI applications, imposing restrictions and safeguards proportionate to the potential societal impact.

The Act adopts a nuanced regulatory approach centred on these core principles:

• Risk-Based Categorization: AI products and services are classified according to their associated risk levels.
• Prohibited Practices: The Act bans AI applications deemed to pose an unacceptable risk to fundamental rights and safety, including social scoring systems, certain forms of biometric identification, and predictive policing tools.
• Transparency and Accountability: Developers of generative AI models must disclose extensive information on datasets used for training, ensuring compliance with copyright protections and necessitating clear labelling of AI-influenced content. Companies are further obligated to report serious incidents related to their AI systems.
• Global Impact: The Artificial Intelligence Act is anticipated to become a global benchmark, influencing the development of AI regulatory frameworks worldwide. China has introduced similar governance initiatives, while various nations in Asia and South America are taking steps toward implementing AI safeguards.

The centrepiece of the AI Act is a risk-tiered classification system:

• Minimal Risk: AI applications such as email spam filters and content recommendation systems fall into this category, facing minimal regulatory oversight.
• High-Risk: AI systems deployed in critical sectors like healthcare or infrastructure are subject to stringent requirements, including pre-market risk assessments, human oversight, and robust documentation.
• Unacceptable Risk: The Act prohibits certain AI practices deemed fundamentally incompatible with EU values. These include government-operated social scoring systems, real-time biometric identification in public spaces, and AI-driven predictive policing.

Companies found in violation of the Act face substantial fines, potentially reaching 7% of their global revenue.

The Act’s provisions are expected to become enforceable in stages following formal approval by EU member states. Fines for non-compliance can reach up to 7% of a company’s global revenue. Individual countries within the EU will be responsible for establishing national AI regulatory bodies.

The passage of the Artificial Intelligence Act marks a watershed moment in the responsible development and use of AI technologies. It establishes a blueprint for balancing innovation with ethical considerations, aiming to protect fundamental rights while fostering a trustworthy AI ecosystem. Continued refinement of the legislation will likely occur in the coming years to adapt to the rapidly evolving field of AI.