1. Methodology

This section details the steps taken by our team to identify, screen, and classify AI soft law programs. The methodology was heavily inspired by prior research aimed at characterizing the state of AI governance [1], [2]. It begins with a description of our multi-pronged strategy to identify a list of candidate programs. The second section details the screening of programs and our approach to classifying them with up to 107 variables and themes. The last section cautions readers on this project’s limitations. Although precautions were taken to guarantee the database’s reliability, we realize that areas of opportunity will be found within our methodology.