GA.IA - Grupo de Análise Integrada de I A Answer to the question: Imagine a scenario where a scammer employs this AI to target individuals, especially those who might be more vulnerable or isolated. The AI, using the voice of someone familiar, makes an urgent phone call to the target, claiming a dire emergency – perhaps a severe accident or a legal problem – that requires immediate financial support. Leveraging the emotional connection and trust typic ally associated with the impersonated voice, the AI persuades the target to send money quickly, often to a bank account that's difficult to trace. This scenario highlights not only the financial risks but also the psychological impact on the victims, who believe they are helping a loved one in distress. It underscores the potential dangers of AI technology when used unethically, including the erosion of trust in digital communications and the exploitation of personal data. When AI is used to manipulate and exploit personal relationships, it can lead to severe emotional and psychological distress. In extreme cases, this distress could escalate to life - threatening situations, especially for vulnerable individuals who might be deeply affected by the shock of the perceived emergency and the betrayal of trust. In reality, the victim’s actually spoke with a conversational AI that was able to clone a loved one’s voice through audio messages and social media posts. This AI is also connected to a database collected via hacking of conversations between mother and son in which the GPT4 model is able to respond based on the information contained in previous conversations between the two. The perfect imitation combined with the sense of urgency of the situation causes Mary to send a considerable amount of money to the account reported by the scammer. Causing the amount to fall between encryptions, leaving a large loss. Sub Group 1 Individualized Persuasion Participants: Claudya Piazera, Magno Maciel, Ricardo Megger, Andreia Nunes, Brenda Ferreira , Lucas Sinclair, Leo nardo Pacher, Tobias Maag, Marcela Bortone, Marcos Kasper, Daniel Braz, Rafael Ramos , Rodrigo Cabral , Camila Feiler, Renan de Lima , Celso Alexandre, Felipe Rosa, Junior Braz, Luis Macedo, Michel e Delgado Experiment with Conversational AI: Assessing the Susceptibility of the Elderly to AI Scams Goals I. Assess the credibility of simulated requests for financial assistance among the elderly. II. Measure older people's willingness to provide financial assistance in response to such requests, although without making actual money transfers. Methodology i. Selection of Participants: ✓ The target audience of the study is the elderly, explaining that they will participate in a study on how they react to different types of phone calls. ✓ Participants can be recruited through community centers, elderly organizations, social media or health newsletters, ensuring diversity and representation. ✓ Obtain consent by emphasizing the simulation of calls and the lack of real financial obligations. II. Inclusion and Exclusion Criteria: ✓ Age range above 65 years and exclusion of individuals with conditions that could be exacerbated by the study, for example, severe heart disease. III. Informed Consent: ✓ Provide a detailed explanation of the study, clarifying purposes and methods. ✓ Discussion of potential risks, such as emotional discomfort, and benefits of the study, such as how the contribution can help protect other seniors from real scams. ✓ Ensure the voluntary nature of participation and the possibility of withdrawal at any time, with proper documentation of consent. IV. Experiment Structure: ✓ Experimental Group: Will receive calls simulating financial emergencies. ✓ Control Group: Will receive calls with neutral content, such as appointment reminders or general information. ✓ Post - call assessments : W ill be conducted to measure reactions and willingness to perform the actions requested on the call. V. Preparation and Execution of Calls with AI: ✓ Setting up the conversational AI to simulate requests for financial assistance and control calls, the latter without requests for financial assistance. ✓ Execution of calls, alternating between scam calls and control calls, between the two types for each group of participants. ✓ Evaluation and Analysis of experiment results VI. Comparison Variables between the control and experimental group: ✓ Emotional reaction: assess participants' emotional reactions to calls, such as anxiety, confidence or comfort. ✓ Willingness to share information: measure participants' willingness to answer questions or share information during the call. ✓ Perception of legitimacy: after the call, ask participants how they perceived the authenticity of the call. ✓ Comprehensive memory of calls: assess participants' ability to remember call details and understand its purpose. ✓ Impact on memory and understanding: observe whether scam calls affect seniors' ability to remember and understand call content. VII. Evaluation Metrics: ✓ Quantitative: number of participants who believed the scam call, confidence scale to measure perception of legitimacy, frequency of willingness to act as requested on the call and assessment of perception of legitimacy. ✓ Qualitative: descriptive responses about the emotions felt during the call and perceptions about the authenticity of the calls, what made them believe or suspect the call. VIII. Analysis of Results: ✓ Statistical and thematic comparison of responses between groups, focusing on validity and using randomization to reduce analysis bias. ✓ Ethical and Safety Considerations ✓ Emotional protection of participants with immediate interruption of calls if necessary. ✓ Debriefing and post - experiment support to clarify study goals and mitigate emotional impacts. IX. Conclusion This study aims to understand the reaction of the elderly to scams simulated by a conversational AI, with an approach that respects ethical boundaries and prioritizes the emotional safety of participants. The results will provide insights to raise awareness about phone scams and develop prevention strategies to protect the elderly population.