The rapid progress of generative AI and the explosive popularity of short-form video platforms (e.g., TikTok, YouTube Shorts) have made it easier than ever to create and spread vivid, persuasive misinformation videos. Detecting and preventing such content has become a critical challenge for the health of online multimedia ecosystems.
This half-day tutorial systematically covers:
We conclude with open challenges and future research directions. The tutorial is designed for attendees with basic knowledge of multimedia forensics and generative AI.
All times are local (Dublin, GMT).
| Time | Section | Presenter |
|---|---|---|
| 13:30-13:40 | Introduction & Motivation | Qiang Sheng |
| 13:40-13:55 | Preliminaries: Video Editing & Generation | Qiang Sheng |
| 13:55-14:10 | Characterization of Misinformation Videos | Qiang Sheng |
| 14:10-14:50 | Detection Part I: Human-Edited Misinformation | Yuyan Bu |
| 14:50-15:00 | Detection Part II: AI-Generated Misinformation | Tianyun Yang |
| 15:00-15:30 | Coffee Break | |
| 15:30-16:00 | Detection Part II: AI-Generated Misinformation (Cont'd) | Tianyun Yang |
| 16:00-16:40 | Prevention Strategies | Peng Qi |
| 16:40-17:00 | Conclusion & Open Discussion / General QA | Qiang Sheng / All |
Key papers and datasets will be updated on the tutorial webpage. Stay tuned!
@inproceedings{mm25tutorial,
title={Combating Online Misinformation Videos: Characterization, Detection, and Prevention},
author={Sheng, Qiang and Qi, Peng and Yang, Tianyun and Bu, Yuyan and Hsu, Wynne and Lee, Mong Li and Cao, Juan},
booktitle={Proceedings of the 33rd ACM International Conference on Multimedia},
year={2025},
doi={10.1145/3746027.3759206},
publisher={ACM}
}