Recently, the significant increase in special fraud cases has highlighted the critical need to understand the psychological and physiological characteristics of victims to enhance prevention. This research aimed to shed light on these aspects, particularly among elderly individuals, by reproducing simulated special fraud scenarios. The objectives were to identify changes in physiological indicators, such as heart rate and respiration rate, and psychological states like tension and confusion, and subsequently, to unravel the mechanisms by which elderly individuals fall prey to fraud. A deception-based experiment was conducted involving 28 elderly participants (mean age 74.4 years). Participants were first exposed to a "daily conversation" audio, followed by a "fraudulent conversation" that offered increased rewards in exchange for personal information for the experimental group. During these interactions, physiological responses were measured, and psychological states were assessed using the Profile of Mood States 2 (POMS2) questionnaire. We also recorded whether participants were "deceived," i.e., whether they were willing to provide personal information for the promised reward. Statistical analysis revealed that the experimental group exhibited significantly higher heart rates and a tendency towards increased tension during the simulated fraud. Further analysis showed that deceived individuals initially experienced agitation, marked by elevated heart rates, but later showed lower heart rates and reduced tension as they seemingly came to trust the fraudulent claims. Applying this detailed understanding of the principles of individual deception, we developed a generative AI-powered dialogue training system for preventing specialized fraud.