FakeShield: An AI-Powered Multi-Signal Fake News Detection Platform

This paper has presented FakeShield, a full-stack AI-powered fake news detection platform demonstrating that principled software engineering can elevate a student ML project to production-grade quality. The system integrates TF-IDF + Logistic Regression classification, multi-signal credibility scoring, live news aggregation, Wikipedia fact-verification, JWT-authenticated user accounts, and persistent analysis history within a cohesive FastAPI + React architecture. A systematic black-book audit identified six security findings requiring immediate remediation, with particular urgency around exposed environment credentials, absent HTTPS, and a non-functional biometric authentication feature that creates false user trust. The proposed six-week roadmap provides a concrete path from the current 68% completeness rating to a deployment-ready, portfolio-grade product. Future work will focus on fine-tuning DistilBERT to address the semantic limitations of the current linear classifier, implementing a Redis caching layer to improve scalability, and extending the platform to support Indian-language content through XLM-R multilingual models and a WhatsApp bot interface—addressing   a  high-impact  real-world misinformation vector in the Indian context.

[1]   V. Vosoughi, D. Roy, and S. Aral, "The spread of true and false news online," Science, vol. 359, no. 6380, pp. 1146–1151, 2018. [2]   S. Sharma, "Misinformation and WhatsApp in India: Implications and policy responses," Observer Research Foundation, Tech. Rep., 2019. [3]   Y. Liu et al., "RoBERTa: A robustly optimized BERT pretraining approach," arXiv preprint arXiv:1907.11692, 2019. [4]   V. Pérez-Rosas, B. Kleinberg, A. Lefevre, and R. Mihalcea, "Automatic detection of fake news," in Proc. 27th Int. Conf. Comput. Linguistics (COLING), 2018, pp. 3391–3401. [5]   F. Ahmed and M. TrabelsiEl, "Detecting opinion spams and fake news using text classification," Security Commun. Netw., vol. 2017, Article ID 9573262, 2017. [6]   W. Yang, Z. Shi, and X. Luo, "Fake news detection with BERT pre-training," in Proc. Int. Joint Conf. Neural Netw. (IJCNN), 2019, pp. 1–8. [7]   A. Conneau et al., "Unsupervised cross-lingual representation learning at scale," in Proc. ACL, 2020, pp. 8440–8451. [8]   Y. Bian et al., "Rumor detection on social media with bi-directional graph convolutional networks," in Proc. AAAI, vol. 34, 2020, pp. 549–556. [9]   K. Popat, S. Mukherjee, A. Yates, and G. Weikum, "CredEye: A credibility lens for analyzing and explaining misinformation," in Proc. WWW Companion, 2018, pp. 155–158. [10] J. Thorne et al., "FEVER: A large-scale dataset for fact extraction and VERification," in Proc. NAACL, 2018, pp. 809–819. [11] V. Sanh, L. Debut, J. Chaumond, and T. Wolf, "DistilBERT, a distilled version of BERT: Smaller, faster, cheaper and lighter," arXiv preprint arXiv:1910.01108, 2019.