A Real-Time Multilingual Fake News Detection

Abstract

The rapid expansion of information across the internet has made it increasingly difficult for users to differentiate between credible and misleading news. Conventional approaches to fake news detection typically rely on rule-based techniques or supervised machine learning models. While these methods can perform effectively on specific datasets, they often struggle to generalize across diverse contexts and are limited in their ability to identify newly emerging misinformation, particularly in multilingual environments. The proposed system addresses these challenges by analyzing multiple content modalities, including textual data, URLs, and images, to better understand the underlying narrative of news items. It verifies claims by comparing them against reliable and authoritative sources during content processing. In cases where no prior fact-check exists, the system employs advanced AI-driven reasoning mechanisms to assess the credibility of the information. Recognized for its capability to identify and mitigate misinformation, the framework is adaptable across various languages and digital platforms. This paper presents a detailed overview of the system’s architecture, operational workflow, implementation strategy, and evaluation results. It highlights how the integration of large language models with robust fact-verification techniques significantly enhances the accuracy and reliability of fake news detection.

Introduction

The text discusses the growing problem of misinformation in the digital age, where false content spreads quickly through online platforms, influencing public opinion and potentially causing social and political instability. Existing detection systems based on traditional machine learning struggle to adapt to new forms of misinformation and often lack support for multiple languages, limiting their effectiveness in diverse regions.

To overcome these limitations, the proposed system introduces a real-time, AI-driven fake news detection framework that supports multilingual analysis. It uses advanced language models like Gemini API to understand context, translate content, and extract factual claims. The system also integrates external verification tools such as Google Fact Check Tools and NewsAPI to improve accuracy.

The architecture is modular, handling inputs like text, URLs, and images (via OCR). It processes data through steps including language detection, semantic analysis, claim extraction, and a dual verification approach—using both external fact-checking sources and AI-based reasoning when no prior verification exists. The final output classifies content as true, false, or misleading, along with clear explanations.

Implementation is done using Node.js, ensuring efficient API communication and real-time processing. The system performs well across multiple languages (100+), provides accurate results, and explains its decisions clearly. However, it may face issues like dependency on external APIs, occasional misclassification of unclear or satirical content, and minor inconsistencies.

Conclusion

This study demonstrates that integrating LLM-based reasoning with real-time fact-checking sources offers an effective and scalable method for detecting misinformation. With features such as multilingual understanding, claim identification, contextual evaluation, and evidence verification, the system remains highly adaptable to evolving forms of disinformation. Several enhancements could further strengthen the system. These include developing a browser extension for instant fact-checking, enabling integration with messaging platforms like WhatsApp, adding capabilities to detect deepfake images, and introducing voice-enabled fact-checking for hands-free interaction. Overall, the system stands out for its adaptability and multilingual capabilities, making it particularly valuable in regions with linguistic diversity. Its ability to explain decisions further enhances user confidence and usability in real-world scenarios.

References

[1] 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. [2] X. Zhou and R. Zafarani, “A Survey of Fake News: Fundamental Theories, Detection Methods, and Opportunities,” ACM Computing Surveys, vol. 53, no. 5, pp. 1–40, 2020. [3] W. Y. Wang, “‘Liar, Liar Pants on Fire’: A New Benchmark Dataset for Fake News Detection,” in Proc. 55th Annual Meeting of the Association for Computational Linguistics (ACL), 2017. [4] A. Vaswani et al., “Attention Is All You Need,” in Proc. 31st Conference on Neural Information Processing Systems (NeurIPS), 2017. [5] Google, “Gemini API Documentation,” 2024. [Online]. Available: https://ai.google.dev [6] Google, “Fact Check Tools API,” 2023. [Online]. Available: https://developers.google.com/fact-check [7] BOOMLive, “Fact-Check Articles Database.” [Online]. Available: https://www.boomlive.in [8] AltNews, “Fact-Checking Reports.” [Online]. Available: https://www.altnews.in [9] NewsAPI, “News Data Aggregation API Documentation.” [Online]. Available: https://newsapi.org [10] S. Kumar and N. Shah, “False Information on Web and Social Media: A Survey,” arXiv preprint, arXiv:1804.08559, 2018.

Copyright

Copyright © 2026 Ojaswi Bangre, Tanvir Malnas, Bhuwan Kanampalliwar, Ritesh Meshram, Tushar Pawar, Mahendra Patil, Sushama Telrandhe. 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.