Anti-scam Application for Multi-level Marketing and ConferencingUsing Artificial Intelligence

Le Trung Thuc; Nguyen Anh Quan; Hoang Manh Cuong; Ngo Thuy Giang; Nguyen Huu Phuong

doi:10.65153/efccc665

Các tác giả

Thuc Le Trung Trường Đại học Công nghệ Đông Á Tác giả
Nguyen Anh Quan Faculty of Information Technology, East Asia University of Technology Tác giả
Hoang Manh Cuong Faculty of Information Technology, East Asia University of Technology Tác giả
Ngo Thuy Giang Faculty of Information Technology, East Asia University of Technology Tác giả
Nguyen Huu Phuong Faculty of Information Technology, East Asia University of Technology Tác giả

DOI:

https://doi.org/10.65153/efccc665

Từ khóa:

Artificial intelligence, Cheat, Fraud detection, Small Language Models (SLMs)

Tóm tắt

The rapid growth of online multi-level marketing (MLM) and virtual conferencing platforms has increased the risk of fraudulent activities. Scammers exploit persuasive communication and social engineering techniques to deceive users, causing financial losses and reducing trust. This paper proposes an AI-based anti-scam application for detecting fraudulent behaviors in MLM and online conferencing environments. The system integrates Small Language Models (SLMs) with Retrieval-Augmented Generation (RAG) to provide efficient, context-aware scam detection in real time. By combining retrieved knowledge from scam databases and regulatory sources with live text or speech analysis, the system identifies deceptive patterns such as exaggerated profit claims and coercive recruitment language. Experimental results demonstrate promising detection accuracy with low latency suitable for live conferencing applications. The proposed approach is scalable, adaptable, and capable of integration with existing online meeting and MLM management platforms, contributing to improved security and transparency in digital communication environments.