Initial Evidence for Inoculation Effects From Daily Rumor Alerts: A Six-Day Randomized Controlled Trial

We examined whether a chatbot-based rumor alert system providing daily periodic alerts over a six-day period could generate inoculation effects against misinformation. In a randomized controlled experiment (N= 56), participants in the experimental group received daily rumor alerts over a six-day period via the LINE messaging app, while the control group read government-published media literacy materials. Pre- and post-intervention questionnaires measured credibility ratings, sharing motivation, confidence, and media literacy competencies for both rumor headlines and true headlines not presented during the intervention. Results showed that the experimental group exhibited significantly lower credibility ratings for both rumor headlines (p= 0.0012, r= 0.64) and true headlines (p= 0.0009, r= 0.65) compared to pre-intervention baselines. Sharing motivation also decreased significantly for both categories (p¡0.05, r= 0.50-0.55). These findings indicate successful inoculation transfer to unpresented content, demonstrating that exposure to specific rumors can promote general vigilance. However, the parallel reduction in credibility for true information reveals an overcorrection effect—a critical consideration for system design. We discuss mechanisms underlying this dual impact and propose system modifications (source credibility indicators, differential alerting strategies) to maintain protection against misinformation while preserving appropriate trust in legitimate information sources. This work provides a proof-of-concept for periodic intervention through automated messaging, demonstrating that passive, periodic inoculation can sustain rumor resistance while identifying design trade-offs requiring careful calibration in practical implementations.