EVID-Bench supplies 222 videos across nine manipulation types in three categories and shows that frontier multimodal models reach at most 61.43% point-level accuracy when forced to use web search to identify false information.
Beyond closed-pool video retrieval: A benchmark and agent framework for real-world video search and moment localization.arXiv preprint arXiv:2602.10159, 2026
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
When Seeing Is Not Believing -- A Benchmark for Search-Grounded Video Misinformation Detection
EVID-Bench supplies 222 videos across nine manipulation types in three categories and shows that frontier multimodal models reach at most 61.43% point-level accuracy when forced to use web search to identify false information.