Provenance information reduces intent to share subtly manipulated images

Abstract

Despite the important role images play in communication, most misinformation research focuses on textual material. Here, we add to a growing body of evidence that finds internet users demonstrate a poor ability to discern manipulated images from real ones, suggesting a need for better tools to help users investigate image authenticity. One such approach would be to make the same processes used by professional fact checkers scalable and accessible to all internet users. We designed and tested mocks for a hypothetical online tool that would conduct an automatic “reverse image search” to identify visually similar images that were posted earlier with less digital manipulation. We find that such a tool might reduce users’ willingness to share images with manipulations that are especially difficult to discern.