The paradox of AI content labeling: how clarity influences information avoidance via cognitive dissonance on social platforms.
Level 2 - randomized trial
Two online randomized controlled experiments evaluating user responses across conditions
PubMed 41884546 · doi:10.3389/fpsyg.2026.1751670
What was done
Two online experimental studies (total N = 760) evaluated how three AI disclosure labeling conditions (clear, ambiguous, and no label) influenced user information avoidance within simulated social media environments (Bilibili and TikTok). The authors assessed cognitive dissonance as a potential mediator, as well as the moderating roles of label-content congruence and thematic relevance.
What was found
Ambiguous AI labels significantly increased information avoidance compared with clear labels or no labels. Cognitive dissonance was identified as a key mediator, leading to user discomfort and subsequent disengagement. Moderating effects were observed for label-content congruence and thematic relevance. The abstract reports no numerical effect sizes, group counts, or test statistics.
Why it matters
Attempting to provide transparency through vague or ambiguous AI disclosure labels can backfire, provoking cognitive dissonance that drives users to avoid content altogether.
Limits
The abstract provides no numerical data, confidence intervals, or exact statistical values. The study relied on simulated rather than naturalistic social media feeds, and the long-term impact on user behavior was not assessed.