Experiencer, Helper, or Observer: Online Fraud Intervention for Older Adults Through Role-based Simulation

TL;DR

ROLESafe introduces a multi-role, LLM-assisted online fraud education intervention for older adults, comparing Experiencer, Helper, and Observer roles against static materials. In a Chinese sample (n=144), Experiencer and Helper roles improved online fraud cue identification versus control, with Helpers outperforming Observers, though follow-up effects were not replicated due to reduced sample size. The study provides empirical support for multi-perspective, cue-based training and discusses design, safety, and generalizability considerations for AI-enabled fraud education targeting older adults. It also highlights opportunities to extend contexts, roles, and longitudinal evaluation, while noting cultural limitations and the need for safeguards in large-scale deployment.

Abstract

Online fraud is a critical global threat that disproportionately targets older adults. Prior anti-fraud education for older adults has largely relied on static, traditional instruction that limits engagement and real-world transfer, whereas role-based simulation offers realistic yet low-risk opportunities for practice. Moreover, most interventions situate learners as victims, overlooking that fraud encounters often involve multiple roles, such as bystanders who witness scams and helpers who support victims. To address this gap, we developed ROLESafe, an anti-fraud educational intervention in which older adults learn through different learning roles, including Experiencer (experiencing fraud), Helper (assisting a victim), and Observer (witnessing fraud). In a between-subjects study with 144 older adults in China, we found that the Experiencer and Helper roles significantly improved participants' ability to identify online fraud. These findings highlight the promise of role-based, multi-perspective simulations for enhancing fraud awareness among older adults and provide design implications for future anti-fraud education.

Figures (14)

• Figure 1: Study overview. The blue sections illustrate the overall study design. The green sections depict the study procedure for an individual participant. Note: participants were randomly assigned to begin with either the Finance or Health context, and the A/B cue sets were also randomly assigned.
• Figure 2: Interaction examples in the learning phase between users and LLM (taking the finance B context as an example). (a) An authentic conversation between an experiencer-group user and the LLM. (b) An authentic conversation between a helper-group user and the LLM. Colored highlights indicate how different fraud cues are manifested by the model.
• Figure 3: An illustration of a fraud cue identification test within a finance context. It demonstrates how specific fraud cues (left) correspond to suspicious elements in a fraud test case (right).
• Figure 4: Descriptive comparisons of effectiveness across groups.
• Figure 5: Descriptive comparisons of user experience across groups.