ABSTRACT Genetics research, like research in sociology and anthropology, creates risks for groups from which research subjects are drawn. This paper considers what sort of protection for groups from the risks of genetics research should be provided and by whom. The paper categorizes harms by distinguishing process‐related from outcome‐related harms and by distinguishing two kinds of group harms. It argues that calls for community engagement are justified with respect to some kinds of harms, but not with respect to others; and (...) it cautions that community engagement may itself be harmful. (shrink)