Fabricated news, known more commonly as fake news, has become a trending issue as its prevalence in media continues to rise. Fake news is characterized as any misinformation that poses as a truthful reporting of news to intentionally mislead people. It is often spread through social media or untrustworthy websites which hold no accountability for how factual the information presented is. This misinformation poses a significant detriment to society as it causes both confusion and lack of trust in the news. To combat this, techniques from data science can be implemented to assist in the identification of fake news. From the results of this analysis, trends can be identified, and predictions can be made, including whether any given news article is legitimate or not. This research applied a Gated Recurrent Unit (GRU) neural network algorithm to develop 3 models utilizing 2 datasets and a third combined dataset to determine the effectiveness of the machine learning algorithm in detecting fake news as well as which of the training sets led to the most reliable performance. From the results, the model developed with the third dataset had the best overall performance with an average 96% accuracy across all datasets, dataset 2’s model performed the best on its own dataset but poorly on the other datasets, and dataset 1’s model’s performance fell in between. This indicates that the algorithm can be effectively used to detect fabricated news but needs a variety of data to train on to be useful.