This study made attempts to discuss and provide a comparison of effectiveness of various machine learning and deep learning algorithms in performing a binary classification flagging of false news, with the intent of making the most accurate and precise predictions. Variety of Models like ANN, CNN, RNN, BLSTM, LSTM, LR, RF, GB, XGB, LGBM, KNN, MLP and ET models were trained and tested for the data set. Evaluation Matrices are fairly standard measures of assessment that were used to test their efficacy. The results showed as ensemble algorithms (Extra Trees, Random Forest, LightGBM) and algorithm of deep learning (MLP) could outperform the other algorithms, the highest results given by the Extra Trees were obtained for the following metrics as Accuracy about 99%, precision, recall and F1-score nearly 99% if were combined with the minimum value of the error rate. On the other hand, CNN, RNN, XGBoost, Gradient Boosting and KNN relatively had less performance, which put its presence to sensitivity towards the dataset attributes. The research goes on to propose that ensemble learning strategy and deep multilayer or optimal accuracy and resilience comes is the best for the binary classification issues. Future endeavors should include extending evaluation to more datasets, larger and more diverse datasets and include explainable AI mechanisms for better model interpretability and exploration of hybrid framework that combines deep learning and ensemble methods for better predictive performance and practical applicability. On top of this, it is possible that by researching Federated and Transfer Learning strategies, you will have an increased speed of the generalizing across domains.