Privacy Concerns in Consumer Adoption of AI Agents

Artificial intelligence (AI) enabled agents such as Apple’s Siri, Amazon’s Alexa, and Google’s Assistant are increasingly important parts of many consumers’ digital lives. According to a recent report from Salesforce, consumers are actively expecting AI agents to assist them with daily tasks. (Salesforce, 2025). Because of the growing importance of AI agents, information systems (IS) researchers should engage in studies that investigate factors that affect how consumers use these agents. Based on prior literature regarding another transformative technology application, electronic commerce, we anticipate that privacy concerns are a potential barrier to consumers’ adoption of AI agents. Our research objective, therefore, is to understand how consumers’ privacy concerns affect their AI agent adoption decisions. To investigate this issue, we synthesize Smith et al.’s (1996) concern for information privacy (CFIP) construct with a widely used theory of technology acceptance, Venkatesh et al.’s (2003) unified model of acceptance and use of technology (UTAUT) (Venkatesh et al., 2003). CFIP consists of four subdimensions: collection, unauthorized secondary use, improper access, and errors. Collectively, these factors shape consumers’ concerns regarding organizational information privacy practices (Smith et al., 1996). UTAUT’s original incarnation includes four main factors that influence technology acceptance: performance expectancy, effort expectancy, social influence, and facilitating conditions. Our overarching theoretical perspective comes from rational choice theory. We treat adoption as a form of decision or choice. Rational choice theory posits that individuals make choices based on an evaluation of the relative costs (including risks) and benefits of each alternative. CFIP provides one form of cost. UTAUT provides a comprehensive model of benefits and additional costs. By synthesizing CFIP and UTAUT we can gain a more comprehensive understanding of how privacy concerns affect adoption decisions within a comprehensive nomological network. We will develop a research model that positions CFIP within UTAUT. We will then conduct a survey of consumers to gather data regarding their beliefs and concerns related to the adoption of AI agents. Data will be analyzed using appropriate statistical methodologies. The resulting findings will offer contributions to both theory and practice.