Owing to ambient intelligence and context awareness in nowadays environments, the provided services become more pervasive and personalized according to the user's profile. With the growing healthcare and wellbeing context aware applications, modeling security policies become an important issue in the design of future access control models. This requires rich semantics using ontology modeling for the management of services provided to dependant people. However, current access control models remain unsuitable due to lack of completeness, flexibility and adaptability to the user capability and behavior. In this paper, we propose a novel adaptable access control model (UBC-ACM) and its related architecture (UBC-ACA) in which the security policy is based on the user's behavior and capability to grant a service using any assistive device within intelligent environment. The design of our model is an ontology-learning and evolving security policy for predicting the future actions of dependent people. This is reached by analyzing historical data, contextual data, and user behavior according to the access rules that are used in the inference engine to provide the right service according to the needs of users.