Authors:
Y. Ajitha, P. Sudha, G. Gowthami, C. Shanthini, Mohammad Ayaz Ahmad

Addresses:
Department of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India. Department of Business Administration, Dhaanish Ahmed College of Engineering, Chennai, Tamil Nadu, India. Department of Computer Science, St. Francis De Sales College (Autonomous), Electronic City, Bangalore, Karnataka, India. Department of Computer Science, Bharath Institute of Higher Education and Research, Chennai, Tamil Nadu, India. Department of Mathematics, Physics and Statistics, University of Guyana, Georgetown, Guyana, South America.

Abstract:

GenAI can create original writing, music, movies, and images and collaborate with humans and other AI models. It is one of our most revolutionary creations. Firms, academics, developers, and consumers have profited from increased productivity, innovation, and efficiency. Generative AI has challenges like any disruptive technology.  It increases creativity and efficiency but creates ethical, disinformation, data privacy, and prejudice concerns.  Recently developed AI tool fraudGPT became popular in the dark market where hackers may automate fraud emails, SMS, manufacture fake text messages, mimic a well-known organisation, and make hacking easy. This traps many innocent people in hackers' webs. AI should be used to detect fraudGPT-related phishing, smishing, social engineering, and financial crimes, according to this study. Python, transformer-based NLP model, anomaly detection model, binary classification, BERT, PyTorch, transformers, and sci-kit-learn are used.  The suggested method detects phishing emails with 92% accuracy and transaction fraud with 94% accuracy in trials.  To test the model's resilience, precision (91%), recall (90%), and F1-score (90.5%) were used for phishing identification and precision (93%), recall (92%), and F1-score (92.5%) for transaction fraud detection. Real-time monitoring and AI-based detection proactively take on AI-driven cyber risks. This work improves AI-powered cybersecurity and informs financial institutions and policymakers.  AI researchers, cybersecurity experts, and regulatory bodies must collaborate to combat rogue AI technologies like FraudGPT.

Keywords: Fraudgpt and Phishing Detection; Natural Language Processing; Transformer Models; BERT and AI-Driven Cybercrime; Pytorch and AI Tools; Text-Based; Transaction-Based; Fraudulent Activities.

Received on: 29/06/2024, Revised on: 12/09/2024, Accepted on: 01/12/2024, Published on: 07/12/2024

DOI: 10.69888/FTSTPL.2024.000335

FMDB Transactions on Sustainable Technoprise Letters, 2024 Vol. 2 No. 4, Pages: 200-211

• Views : 86
• Downloads : 10

Download PDF