Crowd sensing is a data acquisition method that leverages the sensing capabilities of mobile devices to accomplish large-scale and complex data collection tasks. In real applications, the sensing data may contain users’ personal privacy information, posing a risk of privacy leakage upon data upload. To protect their privacy, users might alter genuine data or intentionally upload incorrect data, which adversely affects the sustainable development of crowd sensing. However, most existing decentralized privacy protection schemes rely on randomly selecting aggregate users for data processing, which could include malicious participants, thereby threatening the privacy of users involved in the task. Consequently, these schemes fail to effectively address user privacy concerns. To solve this problem, this paper enhances decentralized privacy protection by integrating a trust-based game model and incorporating collaborative filtering to ensure the accuracy and reliability of uploaded data.