Decisions are widely influenced by biases such as loss aversion, the overweighting of losses relative to gains. Loss aversion is a central driver of human behavior, and “is certainly the most significant contribution of psychology to behavioral economics” (Kahneman, 2011, p. 300). However, while much is known about the role of loss aversion in individual decision-making, its role in social interactions remains controversial. Current theories and indirect evidence suggest that individuals will be good at using loss aversion as an influence tactic in strategic interactions (e.g., negotiations). In contrast, across six studies, we show that individuals (n = 3875 managers, policymakers, and students) are strikingly bad at using loss aversion as an influence tactic. Specifically, across 17 cases where we measured whether individuals optimally used loss aversion as an influence tactic, individuals’ performance in terms of optimally using loss aversion ranged from 40% worse than chance (“backfiringly bad” misuse of loss aversion) to only 18% better than chance (relatively bad performance that is closer to random chance than to optimality). For example, when individuals were given three opportunities to use loss aversion as an influence tactic (i.e., deciding whether to frame an outcome as a “gain” or a “loss”), only 10.5% of managers and 19.0% of policymakers used loss aversion optimally across all three opportunities. In exploring why this happens, we show that individuals’ understanding of loss aversion is (also) strikingly bad. Across 8 cases where we measured whether individuals demonstrated correct beliefs about the direction of loss aversion, individuals’ performance in terms of demonstrating correct beliefs ranged from 59% worse than chance (“backfiringly bad” misunderstanding of loss aversion) to only 15% better than chance (relatively bad performance that is closer to random chance than to correctness). These findings suggest that individuals may fail to understand even the most widespread and powerful of biases, and in many cases will fall prey to “metabiases” – attempts to exploit others’ (decision) biases which backfire, thereby creating a new class of second-order (influencer) biases.