<p>Why do people give false confessions under conditions of conversation-based interrogation? There is a common-sense idea that innocent people don’t confess under non-violent interrogation. Yet we are realizing that false confessions are not uncommon. I survey the work of legal and psychological experts explaining the ordinary psychological mechanisms at work in the modern interrogation context that regularly generate false confessions. Looking at modern interrogation techniques, I will show how they manipulate the epistemic agency of suspects in a variety of ways; further, these techniques depend on manipulating the epistemic agency of many others in the system – detectives, lawyers, judges, and juries. I argue that exploring the details of these various manipulations of wider epistemic agency can help us better understand the systems that elicit and over-credit false confessions, and that this work can provide insight into the appropriate epistemic reparations for these harms.</p>