The term “smishing” combines “SMS” and “phishing,” referring to a cybercrime in which fraudsters use fake text messages to steal personal data or money. Smishing has become a widespread threat as cell phone usage continues to rise. While many solutions exist to combat smishing, none provide a foolproof detection method. The proposed work employs GloVe pre-trained word embeddings alongside machine learning and deep learning models. Specifically, it utilizes the 600-dimensional vector file GloVe. 840 B, which comprises 840 billion tokens. The preprocessing steps include tokenizing text using Keras’s Tokenizer with a vocabulary size of 400,000 words, filtering out non-alphanumeric characters, and padding token sequences to ensure uniform length. This research aims to enhance protection against smishing attacks and improve SMS classification accuracy to 98%.