Today users sharing large volume of images through social sites inadvertently become a major problem of maintaining confidentiality. To help the users to control access to their shared content needs some tools. An Adaptive Privacy Policy Prediction (A3P) used in this paper to address the confidentiality problem.A3P system helps the user to compose confidentiality setting of their images by examine the role of social context, image content and metadata these act as a possible indicators of users privacy preferences.A3P system uses the two-level framework according to users available history on the site to determines the best available privacy policy for users images being uploaded. The solution relies on an image classification framework for image categories which may be associated with similar policies, and on an algorithm which predict the policy to automatically generate a policy for each newly uploaded image, also according to user’s social features. The generated policies fallow the evolution of users’ confidentiality attitude.