Between Privacy and Transparency: Exploring the Ways of Communicating Credibility Assessment

Reducing the spread of misinformation remains a complex problem, especially on encrypted social messaging platforms. AI-based fact-checking systems offer a promising alternative to manual verification, making faster and more scalable responses. However, how these systems communicate their findings to users is still an open design problem. Current approaches, such as binary warning labels, often fail to capture more subtle or partially misleading content. At the same time, users’ limited attention and the overwhelming volume of online information constrain how much and what kind of verification feedback can be delivered. This study explores how two key dimensions of feedback source (content or context-base) and granularity (binary vs. fine-grained assessments) affect users’ trust in the system, perceptions of usefulness, and judgments of content accuracy. In a pre-registered online experiment (n = 537), we tested how these design factors influence user responses. We found that credibility heuristics based on both content and context sources of credibility help users make decisions, and that short-heuristic-based explanations are useful to users. In addition, we found that acknowledgement of system certainty about the verdict also helps users to tailor their opinions about the information. Our findings suggest that context-acknowledged short feedback based on heuristics may be promising to support users in assessing misinformation, even on platforms with limited content visibility, such as encrypted messaging apps.