Vehicular Ad Hoc Networks (VANETs) is a promising technology for autonomous driving that provides many benefits to the user’s conveniences by improving road safety and driving comfort. However, Sybil attack is a major threat in VANETs where an attacker uses multiple fake identities to send false messages, disrupting safety-related applications. To overcome this, a novel jelly Fish Optimized deep Learning for Sybil Attack DETECTion (FOSA-DETECT) has been proposed to detect sybil attack in VANETs and raise the traffic efficiency. The proposed FOSA-DETECT method utilizing the Jelly Fish Optimization for feature extraction to enhance classification accuracy. Convolutional Neural Network (CNN) based Long Short-Term Memory (LSTM) classification technique classifies the extracted feature into two classes such as Normal and Attacked. Measures including specialty, F1score (F1S), Accuracy, Precision (PR), Recall (RC) are used to assess the suggested approach. Compared to current models, experimental results using VeReMi datasets show higher accuracy. The accuracy of the FOSA-DETECT in the VeReMi dataset is 1.6%, 2.8%, and 2.1% higher than that of the current MDFD, MA-DQN and I-LeeNet techniques respectively