The field of cybersecurity is rapidly evolving, with a focus on developing innovative solutions to protect vehicle networks and surveillance systems from cyber threats. Recent research has explored the use of graph neural networks, knowledge distillation, and multimodal reasoning to enhance detection accuracy and reduce computational complexity. These advancements have significant implications for the security of in-vehicle communication systems, such as the Controller Area Network (CAN) protocol, as well as surveillance activities that rely on Voice over Internet Protocol (VoIP) technology. Notable papers in this area include KD-GAT, which combines graph attention networks with knowledge distillation for CAN intrusion detection, and TACTIC-GRAPHS, which introduces a framework for tactical behaviour recognition using causal multimodal reasoning. Additionally, the Hierarchical Graph Neural Network for Compressed Speech Steganalysis and GUARD-CAN, a graph-understanding and recurrent architecture for CAN anomaly detection, have demonstrated impressive results in detecting steganographic patterns and anomalies in vehicle networks.