Multimodal LLMs suffer Safety Geometry Collapse from modality-induced drift that reduces refusal separability; ReGap corrects drift at inference time using self-rectification signals to restore safety without retraining.
arXiv preprint arXiv:2509.25175 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.AI 2years
2026 2representative citing papers
citing papers explorer
-
Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction
Multimodal LLMs suffer Safety Geometry Collapse from modality-induced drift that reduces refusal separability; ReGap corrects drift at inference time using self-rectification signals to restore safety without retraining.
- RankGuide: Tensor-Rank-Guided Routing and Steering for Efficient Reasoning