Aim: This proposed research is to develop fake news detection in SMS (Short Message Service)service using the LSTM-RNN model and improve the accuracy with neural networks in contrast to random forest model. Materials and Methods: The LSTM-RNN model is applied on data, which is a text file containing sequences.a collection of words A random forest for predicting the accuracy of fake news that compares two sources. Model of randomforest It has been suggested and developed to have LSTM. The size of the sample. The G Power value of 0.8 was used to calculate the number of people in each category. The precision was excellent. LSTM-RNN (56 percent) was the most effective in spotting bogus news. When compared to randomforest, the least mean error is (43%). Results: The accuracy was maximum in detecting the fake news in social media using LSTM 51% with long short term memory model 40% for the same dataset. Conclusion: The study proves that LSTM exhibits better accuracy than random forest in detecting the fake news on SMS(Short Message Service) service.