Understanding older adults’ sharing intention toward AI-generated health misinformation: a heuristic-systematic and health belief perspective

The rapid development of generative artificial intelligence has changed the production and presentation of health information on social media. AI-generated health misinformation has increasingly appeared on social platforms, including disease prevention and wellness advice presented by pseudo-authoritative sources, as well as health narratives disguised as personal experience shared by pseudo-layperson sources. Presented in low-barrier, highly visual, and context-rich formats, such content enters older adults’ information environments and may be perceived as convenient and accessible health advice.

Grounded in the heuristic–systematic model (HSM) and the health belief model (HBM), this study focused on influenza prevention and treatment and examined AI-fabricated source and message characteristics. A 2 (AI-fabricated source: pseudo-layperson vs. pseudo-authoritative) × 2 (argument strength: weak vs. strong) between-subjects experimental design was employed to investigate older adults’ sharing intention toward AI-generated health misinformation videos. Using a snowball sampling approach, 320 valid responses were collected from older adults who were willing and able to complete an online experiment through social media platforms.

Analysis of covariance (ANCOVA) and PLS-SEM were conducted. The ANCOVA results showed that under the weak argument strength condition, pseudo-authoritative sources elicited higher sharing intention than pseudo-layperson sources. In contrast, under the strong argument strength condition, the difference between the two source types was not significant.

PLS-SEM analysis further showed significant indirect associations linking the two experimental factors to sharing intention through perceived threat and perceived benefit. This study suggests that, among older adults in social media-based online health information environments, sharing intention toward AI-generated health misinformation videos is associated with a combined evaluation of source cues, argument strength, and health beliefs. By integrating source-related cues, message-quality cues, and health-related evaluations into the explanatory framework, this study clarifies direct and indirect associations underlying older adults’ sharing intention toward health misinformation.

These findings provide implications for health misinformation governance, media literacy interventions targeting older adults who participate in social media-based health information environments, and platform responsibility mechanisms in the era of generative AI. The author(s) declare that no financial support was received for the research and/or publication of this article. Chongqing College of Humanities, Science & Technology, Chongqing, China School of Media and Communication, Shenzhen University, Shenzhen, China The authors declare no competing interests.

The overall research project was reviewed and approved by the School of Media and Communication, Shenzhen University, on September 10, 2025, before any participant recruitment or data collection began. The stimulus-material validation pretest, conducted from December 10 to December 17, 2025, and the main experiment, conducted from January 18 to January 24, 2026, were carried out as two phases of the same approved research project. Both phases followed the approved procedures for voluntary participation, informed consent, anonymous data collection, and post-study debriefing.