Neural network-based chatbot for providing information support to businesses from a cybersecurity perspective

DOI: 10.31673/2409-7292.2024.040014

Abstract

Kushnerov O. S., Vadko K. S., Kuzchenko A. O. Neural network-based chatbot for providing information support to businesses from a cybersecurity perspective

The article examines the process of developing an intelligent chatbot for businesses that provides information support taking into account cybersecurity requirements. The emphasis is on the role of chatbots built on the basis of neural networks in the modern business environment. In particular, their ability to automate the processing of client requests, optimize interaction with users, and improve the quality of customer service is studied. The key cyber threats associated with the use of chatbots are analyzed in detail, in particular those described in the OWASP Top 10 for large language models (LLM). Among the main vulnerabilities, the risks of data leakage, functionality abuse, and query manipulation are considered. Protection methods were studied, including user authentication and authorization, end-to-end data encryption, the use of self-destructing messages, and ensuring compliance with GDPR requirements for personal data protection. Special attention was paid to the integration of cyber protection mechanisms into the architecture of chatbots. The importance of combining artificial intelligence technologies and modern cybersecurity tools to increase the resilience of business processes to threats was emphasized. The conclusions noted that the creation of secure chatbots contributes to the efficiency of companies and ensuring a high level of information protection in the context of digital transformation. Such integration allows minimizing risks and expands the possibilities of using chatbots in various industries, which makes them an important tool for ensuring competitiveness in the context of rapid technological change.

Keywords: chatbots, neural networks, cybersecurity, information support, business process automation, cyber threats, large language models (LLM).