Video Misinformation Detection: A Systematic Review of Manipulation Tactics, Datasets and Algorithms
Highlights
- Many papers on video misinformation detection exist, yet no systematic review connects misinformation definitions to model design choices
- Systematically reviewed 51 of over 2,200 retrieved papers, organized by misinformation strategy, dataset & label design, and detection method
- Model architectures are often designed for specific misinformation strategies (e.g., deepfakes, out-of-context footage)…
- …yet most are evaluated on datasets using binary real/fake labels that do not distinguish between strategies
- These binary labels obscure whether the targeted misinformation strategy is actually detected
- Findings motivate strategy-aware video misinformation detection: models and evaluations should be aligned to the same strategy
Misinformation strategies that were found in the literature.
Sankey showing how the misinformation types (left) are conceptually (middle) and technically (right) detected.
Overview of labels used in datasets and detection algorithms across reviewed papers, highlighting the prevalence of binary real/fake labels.
Abstract
Misinformation research has identified diverse strategies through which false or misleading content gains credibility, spreads, and influences audiences. Yet no systematic review has synthesized how computer science research on video misinformation detection defines and engages with different misinformation strategies. We therefore systematically review 51 of over 2,200 retrieved computer science papers, organizing them by misinformation strategy, dataset and label design, and detection method. We find that while model architectures address specific misinformation strategies, prevailing binary real/fake labels obscure whether each strategy is actually detected, limiting the transparency and diagnostic value of current evaluations. These findings motivate strategy-aware video misinformation detection.
BibTeX
@article{braun2026videomisinformation,
title={Video Misinformation Detection: A Systematic Review of Manipulation Tactics, Datasets and Algorithms},
author={Florian Braun and Truc Hoang and Gianluca Demartini and Isao Echizen and Timo Spinde},
journal={ACM Computing Surveys},
note={Under Review},
year={2026},
url={https://media-bias-group.github.io/Video_Misinfo_SLR}
}