International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 13 Issue: 08 | Aug 2026
p-ISSN: 2395-0072
www.irjet.net
AI-Integrated Fake News Detection: A Comprehensive Survey of Methods, Challenges, and Future Directions Dr. Amarsinh Varpe¹, Khan Nishat Anjum Rais², Dharavath Ganesh³, Rayala Avinash⁴ Bade Yaswanth Sri Pavan Kumar⁵ ¹(CSE Department of Sandip university). ²³⁴⁵(Students, CSE Department Of Sandip University) --------------------------------------------------------------------***-------------------------------------------------------------------
Abstract – The rapid growth of social media, online news
spread unintentionally when users share inaccurate information without proper verification [6].
platforms, and digital communication has significantly increased the spread of false, misleading, and fabricated information. Fake news can influence public opinion, create social confusion, damage reputations, and affect critical decision-making processes [1], [2]. Consequently, Artificial Intelligence (AI), Machine Learning (ML), Natural Language Processing (NLP), Deep Learning, Graph Neural Networks, multimodal learning, and Transformer models have become essential techniques for automatic fake news detection. This survey comprehensively reviews 30 research papers covering traditional machine-learning methods, recurrent and convolutional neural networks, graph-based approaches, multimodal models, BERT and other Transformer architectures, multilingual models, explainable AI, and large vision-language models. The reviewed studies are systematically compared according to their methods, datasets, major findings, limitations, and conclusions. The survey identifies important research gaps including dataset bias, domain adaptation, multilingual detection, multimodal misinformation, explainability, early detection, evidence verification, computational cost, and generalization to unseen events. Based on these findings, an AI-Integrated Fake News Detection System is proposed that combines contextual NLP models, evidence-aware verification, explainable AI, and a user-friendly interface to provide more reliable and understandable prediction
The proliferation of fake news has become a critical societal concern, as evidenced by its impact on political elections, public health responses during the COVID-19 pandemic, and social stability worldwide [7], [8]. Studies have shown that fake news spreads significantly faster and reaches more people than truthful information [9]. This phenomenon underscores the urgent need for effective automated detection systems. Produced daily [10]. Automated fake news detection therefore aims to assist users by analyzing multiple signals including news text, source information, social context, propagation patterns, images, and other multimedia Manual fact-checking, while valuable, is time-consuming and difficult to scale to the enormous volume of information content [11]. Earlier research commonly used manually designed linguistic features with classical classifiers such as Support Vector Machine (SVM), Random Forest, Logistic Regression, and Naive Bayes [12]. Later studies adopted deep learning approaches including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), and attention mechanisms [13], [14]. Transformer architectures such as BERT (Bidirectional Encoder Representations from Transformers), RoBERTa, and XLNet have further improved contextual representation and achieved state-of-the-art performance in many natural language understanding tasks [15].
Key Words: Fake News Detection; Artificial Intelligence; Machine Learning; Natural Language Processing; Deep Learning; BERT; Transformer; Graph Neural Network; Multimodal Learning; Explainable AI; Misinformation.
Recent work has moved beyond text-only classification toward graph-based social network analysis, multimodal fusion, multilingual detection, explainable AI, and evidenceaware verification systems [16]. This survey examines 30 representative studies to understand this evolution and identify directions for practical deployment.
1. INTRODUCTION The internet has fundamentally transformed news consumption by enabling information to be created and shared almost instantaneously across global audiences [3]. Social media platforms such as Twitter, Facebook, and WhatsApp further accelerate this process, but they also provide an environment in which false or misleading information can spread rapidly and uncontrollably [4]. Fake news may be intentionally created to deceive users, influence political opinions, obtain financial benefits, or manipulate public discourse [5]. Misinformation can also
© 2026, IRJET
|
Impact Factor value: 8.315
The main contributions of this survey are:
|
A comprehensive review of 30 research papers spanning traditional ML to state-of-the-art deep learning and transformer-based approaches. A comparative analysis of different methodologies, datasets, and reported performance metrics.
ISO 9001:2008 Certified Journal
|
Page 329