The article investigates security control problems of Takagi-Sugeno (T-S) fuzzy cyber-physical systems (CPSs) against cyber-attacks on sensor measurement and actuator input signals. In T-S fuzzy CPSs, premise variables (PVs) of fuzzy controllers may rely on the measurable states. They can be influenced by the sensor attacks, to bring about the degradation of the control performance. Then, a fuzzy cooperative interaction observer is proposed to construct new, reliable, and available PVs as ones of the controllers. On the other hand, the information usage of sensor and actuator attack estimation errors (AEEs) is insufficient in the existing fuzzy observers. Consequently, it is difficult to further improve the estimation accuracy. Different from them, two auxiliary systems containing the AEE information are constructed. They are able to cooperate with the dynamics of observer's estimation errors, so that the AEE information is fully utilized to enhance the reconstruction accuracy of the attack signals. Furthermore, a class of new fuzzy observer-based security control with the reliable, available PVs is given such that the influence of sensor and actuator attacks and unreliable PVs is removed. Finally, the merit of the presented scheme is illustrated by the MATLAB-based simulation and hardware-in-the-loop experiment.