A cyber–physical system (CPS) is often vulnerable to cyber-attacks that can cause immense damage in real-time to the data and operation of the physical system. To prevent CPS from entering an unsafe state of operation due to Man-in-the-Middle (MitM) attack, an adaptive control strategy based on a novel backstepping framework is proposed that is different from the classical -step predictor scheme. The proposed scheme is computationally faster owing to the fact that the scheme is independent of the design of virtual control laws (VCLs) and also has the capability to handle uncertain system dynamics thereby safeguarding the CPSs from MitM attack. The overall control scheme is proved to be stable in the sense of exponential mean square stability. The article discusses computational complexity analysis of the overall closed-loop control. Moreover, to handle the uncertainties arising from False-Data Injection (FDI), a Radial Basis Function Neural Network (RBFNN), initialized by a Sparrow Search Algorithm (SSA), is employed. The efficacy of the proposed neuro-adaptive control scheme is validated by considering a nonlinear model of the DC motor and comparing the results of the proposed scheme with some existing results.