Purpose <p>The past few years have seen a rapid growth of artificial intelligence (AI) in healthcare. However, only a few applications have been deployed into clinical practice due to, among other things, their characteristics that may not be framed into existing regulations. This study synthesizes the topics investigated on the regulation of AI in clinical healthcare and their characterization.</p> Method <p>We conducted a scoping review following the PRISMA-ScR criteria, searching the Web of Science database for in-depth and peer-reviewed scientific studies involving AI regulation in clinical healthcare (<i>n</i> = 83) until 2023. The topics emerged from text were coded using thematic analysis, identifying categories in different levels of granularity around four dimensions (ethical, legal, social, and technological).</p> Results <p>Each dimension presents unique challenges, and the resulting categories reflect key principles that need to be addressed. The most prevalent categories fell within the legal dimension, highlighting issues related to the current medical devices regulation and the need for alternative evaluation processes considering AI characteristics. We also discuss how recent policies from different countries have been defined to deal with this issue.</p> Conclusion <p>Our results show that our categories are aligned with principles used in AI frameworks worldwide. Many countries are attempting to accommodate their regulatory framework to include AI applications into clinical practice, considering data requirements and functional impacts on patients’ lives or public health systems, with issues ranging from human rights violations, professional accountability, physician-patient relationships, and digital platformization. While a lot of initiatives are taking place, very few enforceable regulations are currently put into practice. Existing regulations still need to be developed to account for the unique characteristics of AI technologies.</p>