04:00
2026-09-10
arxiv.org
artificial-intelligence
MotionBlind: Probing the Illusion of Motion Understanding in Video-LLMs
A new arXiv paper (2609.09528v1) introduces MotionBlind, a contrastive benchmark showing that Video-LLMs cannot reliably judge physically grounded motion, with open models sitting near the 6.25% chanc…