Why do citizens remain silent about everyday corruption even when reporting mechanisms exist and institutions appear credible? This paper argues that silence is not apathy or mistrust but an equilibrium response to asymmetric risk. Focusing on harassment bribes in India, we develop a Bayesian game with two-sided private information in which citizens choose whether to report corrupt demands and honest officials decide strategically whether to make them, each acting under uncertainty about the other. Citizens differ in their private cost of reporting, capturing fear of bureaucratic reprisal, delay and social sanction. The model yields a cutoff strategy: only those with sufficiently low fear costs report while the rest pay quietly. When fear dominates expected enforcement benefit, universal silence is stable even under credible institutions. A two-period extension shows how low reporting emboldens official opportunism, generating a self-sustaining silence trap that threshold-crossing reform is required to escape. Calibration against India's Global Corruption Barometer confirms plausible parameter values are consistent with the observed 12 percent reporting rate. The analysis implies enforcement capacity alone cannot dissolve the equilibrium. Reducing the personal cost of speaking up is the binding constraint.