A defining challenge of AI regulation is creating a framework that is comprehensive, but still results in rules that are tailored to the nuances of AI in different applications, such as in educational access, hiring, mortgage pricing, rent setting, or healthcare provisioning. This paper proposes a new regulatory approach—the Critical Algorithmic System Classification, or CASC—to allow federal regulators to flexibly govern algorithms used in critical socioeconomic determinations. The CASC framework will help preserve crucial consumer and civil rights protections in the algorithmic age, with sensible restrictions and without major changes to the structure of the federal government.
