A METHODOLOGY FOR ANALYZING KEY RISKS IN DEPLOYING AN AI ASSISTANT IN E-COMMERCE PRACTICE

The aim of this work is a systematic analysis of risks arising from deploying an AI assistant in online retail processes on marketplaces (Wildberries, Ozon). Risks are identified and classified into the following groups: technological, operational, financial, reputational, and legal. Based on expert assessments and statistical data, a “probability-impact” matrix is constructed, and critical risks are determined. The novelty of the study lies in proposing a methodology for quantitative risk assessment using simulation modeling of AI assistant failures. The outcome is a set of risk management recommendations (preventive measures, monitoring, redundancy) that can substantially reduce the expected damage.