Factors Influencing Users’ Trust in Agentic AI Browsers : A Quantitative Study of Perceived Competence and Privacy & Security Assurance

Sammanfattning

Agentic AI browsers represent an emerging category of Artificial Intelligence (AI). These browsers can take actions on the user's behalf, such as creating travel plans, booking flights, and changing or booking appointments and meetings, including accessing sensitive personal information. While these systems offer convenience to users, they introduce new trust challenges, including privacy and security issues and loss of control. Although agentic AI is developing rapidly, empirical research examining how users form trust in agentic AI browsers remains limited. To address this gap, we examined which factors influence users' trust in agentic AI browsers, with a focus on Perceived Competence (PC) and Privacy & Security Assurance (PSA) as determinants of Overall Trust. A quantitative survey was conducted among participants in Sweden, preceded by a pilot test to refine the instrument prior to the main data collection. Data were validated and analyzed using Cronbach's alpha and Pearson correlation analysis. The results indicated that both PC and PSA demonstrated statistically significant positive correlations with Overall Trust. However, PSA emerged as the predominant driver, exhibiting a stronger association with Overall Trust (r = .758) than PC (r = .667). These findings suggest that users' trust in agentic AI browsers is more strongly shaped by perceptions of Privacy & Security Assurance than Perceived Competence. As AI technology continues to advance faster than our understanding of its human impact, empirical research becomes essential so that the development of agentic systems does not outpace users' ability to trust them.