Empirical evidence suggests that people often confabulate when they are asked about their choices or reasons for action. The implications of these studies are the topic of intense debate in philosophy and the cognitive sciences. An important question in this debate is whether the confabulation studies pose a serious threat to the possibility of self-knowledge. In this paper we are not primarily interested in the consequences of confabulation for self-knowledge. Instead, we focus on a different issue: what confabulation implies for the special status of self-attributions, i.e. first-person authority (FPA). In the first part of the paper, we propose that FPA is based on a capacity for self-regulation. Accordingly, FPA depends on the extent to which we are able to bridge the gap between our sayings and doings by aligning our actions with our avowed self-ascriptions and vice versa. FPA is withheld when we (systematically) fail at such re-alignment. In the second part of the paper, we contrast our view with the accounts of Scaife (Acta Anal 29:469–485, 2014) and Bortolotti (Rev Philos Psychol 9(2):227–249, 2018). We claim (contra Scaife) that the apparent fact that we cannot reliably distinguish, from a first-person perspective, when we are confabulating and when we are not, does not necessarily undermine FPA. We argue (contra Bortolotti) that a systematic failure to align our actions with our self-ascriptions and vice-versa is a genuine threat to FPA. In the last part of the paper, we introduce the concept of self-know-how—the know-how embodied in the way one is disposed to relate to oneself in making sense of oneself with or in the face of others—and briefly explored the importance of diminished or absent self-know-how in clinical cases.