<p>The rapid development of intelligent and connected vehicles has led to large-scale sharing of privacy-sensitive automotive data, raising increasing concerns about privacy protection. Existing data-sharing mechanisms typically rely on static and uniform privacy configurations, which fail to accommodate heterogeneous access risks and personalized privacy requirements, resulting in suboptimal trade-offs between privacy protection and data utility. This paper proposes a risk-aware personalized differential privacy approach for automotive data sharing. The approach integrates risk-adaptive access control with personalized differential privacy, where access risks are quantitatively evaluated and mapped to individualized privacy budgets. These budgets further guide privacy-policy selection and privacy-mechanism configuration in a unified manner. To support common automotive data-analysis tasks, a unified personalized local differential privacy model is established. Within this model, multiple frequency-estimation mechanisms are designed for categorical data under different encoding strategies, including longitudinal privacy protection for repeated collection of static or slowly-changing data. In addition, two methods for numerical data are developed, including a personalized mean estimation method based on adaptive bucketing and a piecewise mechanism. Experimental results demonstrate that the proposed approach significantly improves privacy flexibility and data utility compared with conventional global-budget schemes, validating its effectiveness for privacy-preserving automotive data sharing.</p>