AI-generated deepfakes are disrupting public trust in political figures by producing hyperrealistic but false images and videos. While some deepfakes are humorous, others carry malicious intent, harming the reputations of high-profile individuals, including politicians. An exceptionally damaging form involves pornographic content. It disproportionately targets women and has already impacted figures such as US Representative Alexandria Ocasio-Cortez. Drawing on an observational study and an online experiment, our study demonstrates how these sexually explicit deepfakes affect the public's perception of politicians' credibility and competence. Deepfakes blur the boundary between reality and fiction, making them hard to dismiss even after being exposed as false. This poses a particular risk, as people struggle to update their beliefs once initial (mis)perceptions set in. Hence, seeking to offer a remedy to this vicious phenomenon, we also explore whether preemptive debunking can mitigate its adverse effects. Our findings underscore the serious threat deepfakes pose, especially to women, and aim to develop protective strategies to uphold public trust in the democratic process.