The exponential growth of online information has led to significant information overload, emphasizing the need for efficient recommendation systems. One prominent application is movie recommendation engines, which commonly use content- based filtering due to its simplicity and flexibility. However, these systems often face critical challenges, such as the cold start problem, where limited user interactions hinder accurate recommendations. In this paper, we propose a genre-based movie recommendation system that employs a long short-term memory (LSTM) model to tackle the cold start issue. The proposed approach utilizes genre-based features of movies to capture non-linear relationships and predict user preferences for newly introduced or unrated movies. Using genre labels as a key input, the LSTM model identifies latent patterns and provides personalized recommendations even in the absence of prior user ratings. Experimental results demonstrate that our LSTM-based method effectively addresses the cold start problem and improves the accuracy of recommendation in sparse datasets.