The European Union Enacts the First Comprehensive Artificial Intelligence Law

After considerable debate, on December 8, 2023 the European Union reached a provisional agreement on the first comprehensive AI law. The law defines an AI system as “a machine-based system that [...] infers from the input it receives how to generate outputs such as predictions, content, recommendations, or decisions that can affect physical or virtual environments." Key provisions categorize AI into prohibited, high-risk, limited risk and minimal risk.

Prohibited risk includes systems that manipulate human behavior affecting free will, social scoring based on behavior or personal characteristics, certain predictive policing, emotion-recognition technology in the workplace and schools and remote biometric identification in public spaces, with specific exemptions for law enforcement. High-risk systems include critical infrastructure, medical devices, law enforcement, administration of justice and influencing elections and voter behavior, among others. These systems will be required to implement risk mitigation, quality controls, bias mitigation, human oversight and transparency for users, such as labeling deepfakes, and cybersecurity. General purpose AI systems and foundation models, which pose limited or minimal risk, will require disclosure requirements regarding training data and copyright and intellectual property safeguards. While the EU Parliament and Council negotiators reached a provisional agreement on the text of the AI Act, it must be formerly adopted by the Parliament and European Council, which defines the general political direction and priorities of the EU.

Most countries continue to rely upon existing laws that arguably only apply to certain AI concerns. Only China has enacted any law or policies specific to AI governance. In the U.S., the White House Office of Science and Technology Policy released a Blueprint for an AI Bill of Rights in the fall of 2022, followed by President Biden’s Executive Order on Safe, Secure and Trustworthy Artificial Intelligence in October 2023. Congress has yet to pass comprehensive AI legislation.