In the last decade, cybersecurity in the automotive sector has become a subject of great interest, gaining significant attention from researchers and experts. Each modern vehicle is equipped with plenty of sensors monitoring various functionalities, interconnected through a specific bus implementing the Controller Area Network (CAN) protocol. While CAN offers many advantages, it also poses serious issues that can jeopardize not only vehicle stability but also the safety of the driver and passengers due to its design lacking consideration for security and associated driving risks. In this paper, the usage of an Autoencoder modeled with Long Short-Term Memory (LSTM) for detecting anomalies on the CAN-bus is examined. To accomplish this, professional hardware emulators have been employed to generate a synthetic dataset, which includes both normal and anomaly CAN traffic. The experimental evaluation demonstrated the effectiveness of our approach in detecting anomalies in CAN-bus traffic compared to the current state-of-the-art methods.