DropVLA: An Action-Level Backdoor Attack on Vision-Language-Action Models
arXiv:2510.10932v5 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models map multimodal perception and language instructions to executable robot actions, making them particularly vulnerable to behavioral backdoor manipulation…
Read the full story at arXiv cs.AI ↗
ImpactNotable 45/100
Why it mattersRule-based estimate: event keywords (+18); trust 6/10.
Regions🌐 Global
Published1 d ago (Mon, 14 Sep 2026 04:00:00 GMT)
RetrievedMon, 14 Sep 2026 17:39:54 GMT via rss
ClassifiedMon, 14 Sep 2026 17:40:00 GMT by heuristic
AuthorZonghuan Xu, Jiayu Li, Yunhan Zhao, Xiang Zheng, Xingjun Ma, Yu-Gang Jiang