ENSURING INFORMATION PROTECTION WHEN WORKING WITH MODERN ARTIFICIAL INTELLIGENCE TOOLS

Abstract

The study aims to systematize information security risks arising from the use of modern artificial intelligence tools and to substantiate the feasibility of deploying such tools locally for processing restricted-access information. The methodology combines structural and functional analysis, classification of information flows, comparative analysis of local and cloud deployment models, and assessment of information security threats. The results systematize the types of data provided to AI chatbots, AI workflows, and AI agents in personal and corporate environments and identify risks associated with transferring such data to external services. The analysis shows that local model execution, when combined with appropriate network and administrative controls, can keep data processing within a controlled security perimeter. The study concludes that local AI solutions should be prioritized for confidential information, while recognizing the need for adequate hardware resources, qualified personnel, integration skills, and security procedures.