Authors:
Sahaj Bhandari

Addresses:
Department of Mathematics, The Academy for Mathematics, Science, and Engineering, Rockaway, New Jersey, United States of America.

Abstract:

The introduction of AI language models has led to the production of deep fakes, creating dangerous threats to information integrity as well as public trust. In this paper, a new paradigm of trusted algorithms that are powerful enough to neutralise the threat by detecting original text from toxically produced text is introduced. Our model is a single, multi-dimensional model with the latest sentiment analysis, context-sensitivity, and a new trust-scoring process. To ensure our model is credible, researchers used a large, balanced dataset of human- and AI-generated text from diverse sources, including news articles, social media posts, and financial reports. The data are sourced from publicly accessible sources and artificially created using the latest language models. The primary tools used here are Python and its rich NLP libraries, such as NLTK and spaCy, as well as machine learning libraries such as TensorFlow and PyTorch for testing and implementing models. Experimental results indicate that our algorithms are far more accurate at detecting fraudulent content than conventional methods. The research presents a clear methodology, data, and tools, demonstrating that it provides an efficient solution to one of the largest AI time challenges, thus paving the way for safer and more reliable AI applications.

Keywords: Impostor Content; AI Language Models; Reliable Algorithms; Sentiment Analysis; Contextual Understanding; Artificial Intelligence; Natural Language Processing; Research and Learning.

Received on: 05/01/2025, Revised on: 15/03/2025, Accepted on: 15/05/2025, Published on: 07/12/2025

DOI: 10.69888/FTSIN.2025.000551

FMDB Transactions on Sustainable Intelligent Networks, 2025 Vol. 2 No. 4, Pages: 198-206

• Views : 66
• Downloads : 10

Download PDF