MANETs enable ad hoc wireless networking in unstructured and constantly changing environments. As a result, MANETs face a variety of security challenges associated with their lack of centralized control, unencrypted communication and constant changes in the network topology. The study explores the two major threats to MANETs: routing attacks and issues related to trust. A newly developed framework is introduced that integrates trust-based routing, lightweight encryption and artificial intelligence-powered anomaly detection to improve the security of mobile ad hoc networks. Evaluation of the proposed framework using the UCSD/mobility dataset shows that it achieves better packet delivery rates, lower latency, higher intrusion detection accuracy and less routing overhead compared to SAODV and ARAN. The results emphasize that the proposed framework effectively combats security threats without compromising the performance of the network.