We ask what ethical regulations govern criminal legal system research that is not “biomedical and behavioral research”—including oral history and archival projects, legal work and research, journalistic projects, big data, and multidisciplinary projects—but nonetheless takes place inside the academy? We examine ethical frameworks for research in the social sciences, as well as participatory action, oral history, archival, and data use, attending to how these academic ethical frameworks define risk and how these definitions shape resulting research, in intentional and unintentional ways. Through analysis of examples from each of these frameworks, we argue that efforts to eliminate risk often create other harms, while distracting from more fundamental ethical questions about the well-being of research subjects and data contributors in the criminal legal system. We identify an alternative to risk elimination: risk metabolization, a more collaborative and iterative approach to managing risk in research ethics.