Because European Union has spent a lot of effort on GDPR and tries to help people take back control of personal data, and the scandal of leaking or misusing people's data keep happening on the tech companies, people start to become aware of the importance of personal data ownership and control. In the data era, data analysis on personal data is unavoidable no matter from the angle of the values it brings to us, or the angle of the market size of data economy. To protect people's privacy from personal data analysis, there are two directions to do the privacy-preserving computation: run computation locally, or make the working nodes to compute a function over their inputs while keeping the inputs private. In this talk, we will share the status of contemporary edge AI and privacy-preserving computation technologies: 1. Current status of edge AI technology, and the relationship between edge AI and privacy. 2. The possibility that how can users use external AI services without sending personal data out. 3. The possibility that how can users send personal data to external AI services without leaking personal information.