With the rapid development of the Internet and information technology, China′s digital economy is developing rapidly and has become a new driving force for China′s economic growth. As a mainstream platform for communication in the digital economy, the usefulness and attractiveness of social networking platforms derive from the sharing of data among users. The continuous increase in the number of platforms′ users has brought a large amount of data resources. The continuous upgrading of platforms′ services makes the data information generated by users have strong interest attributes, and it is easy to become the target of hackers or competitors. This has led to frequent leakage of private information, continuous negative news on social network platforms, and increasingly prominent personal privacy and security issues of users. Social network data privacy protection issues have received more and more attention.The current research on social network privacy protection mainly focuses on the willingness and choice, influencing factors and mechanisms of the privacy protection behavior of both platforms and users through empirical and evolutionary game methods. There are few studies on platform and user privacy protection decision in social network environment. Moreover, in the actual social network environment, the privacy protection level of the system is not static. Therefore, this article uses differential game theory as an approach to study the multi-stage decision-making problem of privacy protection between platforms and users in social networks from a dynamic perspective, including how both parties achieve a balance between benefits and costs, and how different factors affect the overall and individual benefits of the platform and users, which makes the research closer to reality and provides a theoretical basis for social network platforms and users to formulate privacy protection policies.In the first part, we start from the perspective of privacy protection decisions and government regulation between social network platforms and users, and constructs a differential game model and assumptions between a single user and a single social platform based on treating both parties as rational “economic man”.In the second part, on the basis of satisfying the maximization of interests, we coordinate the efforts of users to share private information and the investment in privacy protection of the platform. And taking the multi-stage development of social network platforms as the main line, Nash non cooperative game models with independent decision-making between platforms and users, Stackelberg master-slave game models with platform incentives, and cooperative game models with consistent decision-making between platforms and users were constructed, and their optimal strategies and key variables were derived. Then, we compare and analyze the optimal strategies of users and social platforms in three stages: non-cooperative game of independent decision-making, Stackelberg master-slave game under the incentive of social platform, and cooperative game of unanimous decision-making. We also discussed the effects of different parameters on the optimal decision-making and optimal efficiency.In the third part, we validated the effectiveness and rationality of the multi-stage differential game decision-making model in the second part through numerical simulation experiments, and more intuitively compared the differences in system optimal returns, optimal paths, and the effects of different parameters on optimal decisions and benefits under the three stages.In the fourth part, through the above analysis and verification, the following conclusions are drawn: firstly, the platform can significantly improve the willingness of users to share information by sharing the cost of users, but compared with the non-cooperative model, it will not improve the level of privacy protection of the system. Platform incentives can effectively enhance the enthusiasm of users to share personal information. Secondly, Under the cooperation mode of rational distribution of interests, the benefits of both parties, system benefits, and the optimal path of privacy protection are strictly superior to the other two modes, and are Pareto optimal. Thirdly, in the case of government supervision, the cooperative decision-making model among the three models can significantly improve the total social benefit, and the government′s participation helps to improve the overall benefit and maximize the social benefit.Finally, according to the conclusions, relevant management implications are put forward: strengthening the platform data management capability, improving the input and output of privacy protection, improving the platform service and subsidy mechanism, and giving full play to the government′s supervision and guidance role. Based on the different development stages of social network platforms, this paper′s conclusions and management implications provide theoretical reference opinions for regulating the development of China′s social network industry. It has important practical significance for achieving win-win business interests between users and platforms, reshaping users′ data trust, and ensuring the sustainable development of the national digital economy.