A new benchmark dataset and two tasks let researchers test whether AI systems can judge if a news image's recorded location and date match the article, with current chatbots scoring 64-81% on location but 42-58% on date.
This photograph has been altered: Testing the effectiveness of image forensic labeling on news image credibility
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Despite the ubiquity and proliferation of images and videos in online news environments, much of the existing research on misinformation and its correction is solely focused on textual misinformation, and little is known about how ordinary users evaluate fake or manipulated images and the most effective ways to label and correct such falsities. We designed a visual forensic label of image authenticity, Picture-O-Meter, and tested the label's efficacy in relation to its source and placement in an experiment with 2440 participants. Our findings demonstrate that, despite human beings' general inability to detect manipulated images on their own, image forensic labels are an effective tool for counteracting visual misinformation.
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Dataset of News Articles with Provenance Metadata for Media Relevance Assessment
A new benchmark dataset and two tasks let researchers test whether AI systems can judge if a news image's recorded location and date match the article, with current chatbots scoring 64-81% on location but 42-58% on date.