Legal Elements of an AI Regulatory Permit Program

Abstract

The potential far-reaching risks some artificial intelligence (AI) technologies pose to human rights and public interests highlight the importance of lawmakers’ present and future decision about what legal regulatory frameworks they will impose on AI companies. One risk-mitigating regulatory approach lawmakers could choose is permitting, whereby government administrators act as market gatekeepers, allowing AI systems to operate in commerce only if they meet specific technical and other standards, and if their owners demonstrate they can operate their systems within acceptable levels of risk while also maintaining compliance with individualized permit terms and conditions. The key legal elements of a national AI permit program are described here, modeled in part on other regulatory permit and general approval schemes used in the U.S. The legal elements are contrasted with a proposed general approval approach for AI-based medical devices and the European Commission’s proposed regulatory framework for AI. Critical differences between the approaches are considered, which reveal how enforcement mechanisms in particular could make a big difference in AI risk reduction.

Market for artificial intelligence in health care and compensation for medical errors

“…These two elements raise questions about the liability of actors toward patients if a medical error occurs. The challenge lies in the fact that AI systems may be the direct or indirect cause of patient harm 2 as physicians are also expected to exercise control over them 3 (Higgins, 2022). As a consequence, it is relevant to study the division of compensation for harm paid by the producers of AI systems and physicians.…”