We address the proof-based development of (system) models satisfying a security policy. The security policy is expressed in a model called OrBAC, which allows one to state permissions and prohibitions on actions and activities and belongs to the family of role-based access control formalisms. The main question is to validate the link between the security policy expressed in OrBAC and the resulting system; a first abstract B model is derived from the OrBAC specification of the security policy and then the model is refined to introduce properties that can be expressed in OrBAC. The refinement guarantees that the resulting B (system) model satisfies the security policy. We present a generic development of a system with respect to a security policy and it can be instantiated later for a given security policy.