Center for Data Innovation (December 2021)
with Ben Mueller
The EU’s proposed Artificial Intelligence Act (AIA) would create a risk-based framework for regulating AI, with designated “high-risk” sectors subject to a long list of rules that regulate how firms can design, train, and deploy AI systems. The law’s definition of AI is critical: It determines what types of software applications must abide by these obligations. However, the law’s definition of AI is so sweeping that it would regulate a broad array of software, creating significant costs and seriously damaging the European Commission’s Digital Decade ambitions. A narrower definition of an “AI system” will make the AIA more appropriately targeted and, accordingly, less expensive for the European economy.