Machine learning based access control (MLBAC) models are emerging as efficient access control systems today that can make accurate access control decisions while reducing the administrative burden on human administrators. With the emergence of MLBAC models, there is an administrative problem that needs to be addressed, updating and revising access control states in a system, such as revoking a user’s access to an object. This paper focuses on the administration problem of an instance of the MLBAC model, i.e., environment aware deep learning based access control (DLBAC-Env) model. DLBAC-Env is a deep learning model that makes access control decisions based on user, resource, and environment metadata within a system, providing outputs to either allow or deny access control requests. To address the administration problem in this model, we utilize a non-symbolic machine learning approach to incorporate changes in the access control states. Researchers have shown that non-symbolic approaches outperform symbolic ones. Our experimental results for the DLBAC-Env administration problem demonstrate comparable outcomes to previous work, indicating its feasibility and applicability in real-world scenarios.