The Fake News Detection System is a machine learning–based web application designed to analyze online news articles and determine their credibility. The rapid rise of misinformation across social media platforms has created a need for automated tools that assist users in identifying potentially false or misleading news. This project addresses that need by combining Natural Language Processing (NLP) techniques with a supervised machine learning model to classify news as Likely Real or Likely Fake. The system accepts input in two flexible formats — direct text or a URL link to an online article. If a URL is provided, the system automatically extracts the main news content using a web scraping module. The extracted or pasted text is then cleaned, tokenized, lemmatized, and transformed into numerical TF–IDF features. These features are passed into a trained Logistic Regression classifier, which outputs both a predicted label and a confidence percentage. This makes the system not only accurate but also easy to interpret. A simple and user-friendly web interface, developed using HTML, CSS, Bootstrap, and Flask, allows users to interact with the application effortlessly. The backend ensures quick response times, and the overall pipeline—from ingestion to prediction—executes within seconds. The model achieved strong performance on the ISOT Fake News Dataset, with accuracy and F1-score above 90%, demonstrating reliability in distinguishing real news from fabricated or sensational content. This project effectively showcases the power of NLP and machine learning in solving real-world problems. It provides practical value by helping users critically evaluate online information while highlighting the importance of automated misinformation detection tools in today’s digital landscape.