In times of increasing political polarization, articles written in the news tend to become biased, making it more difficult to determine what is true and what is not. In fact, most Americans believe that 62% of news consumed from television and newspaper articles is biased, and they estimate that 80% of news from social media is biased. These biases skew facts in a certain direction to encourage people to make a certain decision. Human analysts require a lot of time to process a given article, meaning they cannot quantify the bias in many articles on the internet. To improve upon this, the goal of my proposed method would be to detect and quantify bias using machine learning. It would be able to process the text of articles and give it a score for the level of bias, as well as indicate which political side it is biased for. Specifically, the model would use recurrent neural networks, in which text is processed in chronological order, such that the interpretation of any word considers the words that precede it. The proposed model would be trained on media already known to be biased, such as articles from news outlets claiming to be advocative rather than objective. Ultimately, citizens would be more cognizant of the level of bias present in their media, so that they could read from more objective sources, or even read an opposing viewpoint.