AI Fake News Detection System

Abstract

The rapid growth of digital media and online news platforms has significantly increased the spread of misinformation and fake news. This paper presents a machine learning-based approach for detecting fake news articles using natural language processing techniques. The proposed system utilizes a labeled dataset of news articles to train a classification model capable of distinguishing between real and fake news. The textual data is preprocessed using techniques such as text cleaning, tokenization, stopword removal, and stemming to improve data quality. Feature extraction is performed using the Term Frequency-Inverse Document Frequency (TF-IDF) method, which converts textual content into numerical vectors. A Logistic Regression algorithm is then applied as the classification model to predict the authenticity of news articles. A user-friendly web interface is developed using Streamlit, allowing users to input news content and receive real-time predictions. The system demonstrates effective performance in identifying misleading information and provides a simple yet practical solution for fake news detection. This approach highlights the potential of machine learning in addressing the challenges of misinformation in the digital era.

Introduction

It explains that the rapid growth of social media has made information easily accessible but has also led to a major rise in misinformation and fake news, which can influence public opinion, politics, and social stability. Because manual fact-checking is too slow for the massive volume of online content, automated detection systems are needed.

Proposed approach:

The system uses a simple and efficient machine learning pipeline:

• Data collection from labeled news datasets
• Text preprocessing (cleaning, lowercasing, stopword removal, stemming)
• Feature extraction using TF-IDF
• Classification model using Logistic Regression
• Web interface built with Streamlit for user interaction

Users can input a news article, and the system predicts whether it is real or fake.

Literature review summary:

Previous research shows that:

• Traditional ML models (SVM, Naïve Bayes, Logistic Regression) work well with good feature engineering
• Deep learning models (CNN, LSTM, BERT) achieve higher accuracy but require more data and computing power
• Hybrid models using content + social context improve performance but are more complex
• TF-IDF + Logistic Regression remains popular due to its simplicity and efficiency

Methodology:

The system follows a standard pipeline:

1. Dataset preparation
2. Text preprocessing
3. TF-IDF vectorization
4. Train-test split
5. Logistic Regression training
6. Prediction and evaluation
7. Web deployment using Streamlit

Results and discussion:

• The model achieves good accuracy in detecting fake news
• Works better on longer, more informative articles
• Struggles with ambiguous or short text
• Cannot verify real-world facts—only relies on learned patterns
• Streamlit interface improves usability
• Performance could be improved using deep learning or larger datasets

Conclusion

The proposed Fake News Detection System demonstrates the effective application of machine learning and natural language processing techniques for identifying misleading news content. By utilizing text preprocessing, TF-IDF vectorization, and Logistic Regression, the system is able to classify news articles as real or fake based on learned textual patterns. The experimental results indicate that the model performs satisfactorily on the given dataset and is capable of providing reliable predictions for well-structured input data. The integration of a Streamlit-based web interface further enhances the usability of the system by allowing users to interact with the model in a simple and efficient manner. However, the system has certain limitations. It relies entirely on historical training data and does not perform real-time fact verification using external sources. As a result, the model may produce incorrect predictions for newly emerging or contextually complex news articles. The performance of the system is also dependent on the quality and diversity of the dataset used for training. Future improvements can include the use of larger and more diverse datasets, integration of real-time news APIs, and implementation of advanced deep learning models such as LSTM and transformer-based architectures. These enhancements can significantly improve the accuracy and robustness of the system. In conclusion, the proposed system provides a practical and efficient approach for fake news detection and serves as a strong foundation for further research and development in this domain. In addition, the proposed system highlights the importance of integrating machine learning techniques into real-world applications for tackling the growing problem of misinformation. As the volume of online content continues to increase, automated systems such as the one presented in this paper can play a crucial role in assisting users to make informed decisions. The current implementation focuses primarily on textual analysis; however, fake news detection can be further enhanced by incorporating additional features such as image verification, source credibility analysis, and user behavior patterns. Combining these features with machine learning models can significantly improve the overall effectiveness of the system. Moreover, the scalability of the system can be improved by deploying it on cloud platforms and integrating it with real-time data streams. This would enable continuous learning and adaptation to new types of fake news, making the system more robust and reliable over time. Overall, the proposed approach demonstrates that even simple and efficient machine learning techniques can provide meaningful results in fake news detection when combined with proper preprocessing and feature extraction methods. This work can serve as a foundation for future advancements in automated misinformation detection systems.

References

[1] H. Allcott and M. Gentzkow, “Social Media and Fake News in the 2016 Election,” Journal of Economic Perspectives, vol. 31, no. 2, pp. 211–236, 2017. [2] K. Shu, A. Sliva, S. Wang, J. Tang, and H. Liu, “Fake News Detection on Social Media: A Data Mining Perspective,” ACM SIGKDD Explorations Newsletter, vol. 19, no. 1, pp. 22–36, 2017. [3] V. L. Rubin, N. Conroy, and Y. Chen, “Towards News Verification: Deception Detection Methods for News Content,” Proceedings of the Hawaii International Conference on System Sciences (HICSS), 2016. [4] N. J. Conroy, V. L. Rubin, and Y. Chen, “Automatic Deception Detection: Methods for Finding Fake News,” Proceedings of the Association for Information Science and Technology, vol. 52, no. 1, pp. 1–4, 2015. [5] W. Y. Wang, “Liar, Liar Pants on Fire: A New Benchmark Dataset for Fake News Detection,” Proceedings of the Annual Meeting of the Association for Computational Linguistics (ACL), 2017. [6] A. Ruchansky, S. Seo, and Y. Liu, “CSI: A Hybrid Deep Model for Fake News Detection,” Proceedings of the ACM Conference on Information and Knowledge Management (CIKM), 2017. [7] S. B. Parikh and P. K. Atrey, “Media-Rich Fake News Detection: A Survey,” IEEE Conference on Multimedia Information Processing and Retrieval (MIPR), 2018. [8] J. Thorne and A. Vlachos, “Automated Fact Checking: Task Formulations, Methods and Future Directions,” Proceedings of COLING, 2018. [9] F. Ahmed, O. A. Abulaish, and A. A. Alzahrani, “Using Machine Learning for Fake News Detection,” International Journal of Computer Applications, vol. 182, no. 1, pp. 1–7, 2018. [10] Scikit-learn Developers, “Scikit-learn: Machine Learning in Python,” Available: https://scikit-learn.org [11] NLTK Project, “Natural Language Toolkit Documentation,” Available: https://www.nltk.org [12] Kaggle, “Fake News Dataset,” Available: https://www.kaggle.com

Copyright

Copyright © 2026 Kashif Ahmad Khan, Ahrar Shah, Syed Mujahid Ali, Mohammad Zahid. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.