In many languages, φ-agreement with a DP is disrupted when that DP is involved in an A'- dependency (e.g., wh-movement, relativization). Since Ouhalla 1993, this phenomenon has been known as anti-agreement. In this paper, we investigate cases of anti-agreement in which an A'- extracted argument triggers anti-agreement not only on its locally φ-agreeing head, but also on a lower head that φ-agrees with a pronoun bound by the extracted argument (bound anti-agreement). Bound anti-agreement has received little attention in previous literature. Crucially, since the bound variable does not itself undergo movement, bound anti-agreement presents a challenge for analyses of anti-agreement that derive the effect through constraints on A'-movement. We argue instead that bound anti-agreement lends strong support to Baier's (2016) featural theory of anti-agreement, in which anti-agreement is the result of a φ-probe agreeing with an A'-operator bearing the feature [wh]. We moreover analyze bound anti-agreement in a parallel fashion to fake indexicals, thus unifying these two seemingly separate phenomena. Bound anti-agreement supports an analysis of fake indexicals in which a variable is 'born' underspecified, but acquires its features from its antecedent in the course of the derivation (e.g. Kratzer 2009). The features of the variable may be realized on the variable itself (as with fake indexicals) or as agreement with the variable (as seen in cases of bound anti-agreement).