In recent years, the application of Artificial Intelligence (AI) in different sectors of society has become one of the most common technologies. As AI becomes a useful tool, incorporating AI in weather forecasts has also started rising. Lightning is one of India’s leading causes of casualty associated with natural disasters. An early warning system for lightning forecast is of utmost importance. However, the atmospheric processes associated with lightning are yet to be fully understood, making it challenging to simulate in a numerical weather prediction (NWP) model for operational purposes. Considering all these factors, in this chapter, we have discussed an AI-based deep learning approach for lightning prediction. This new AI-based lightning prediction system uses the Indian Meteorology Department (IMD) Weather Research and Forecasting (WRF) model at 9 km resolution as input which presently do not have any lightning product. The Indian Institute of Tropical Meteorology (IITM) Lightning Location Network (LLN) observation data is used as target vector to train the model. Due to the high imbalance of rare and severe events (RSEs) like lightning, we have used a novel approach of two-auto encoder-based classification ( \({C}_{2AE}\) ) to train a deep-learning algorithm to correctly encode meteorological features associated with lightning and non-lightning incidents and, when the trained \({C}_{2AE}\) model is tested on a new dataset, it is able to correctly identify lightning events with statistical significance.