Aligned LLMs exhibit Refusal-Escape Directions (RED) that enable refusal-to-answer transitions via input perturbations; these directions decompose exactly into operator-level sources, creating an inherent safety-utility trade-off when trying to eliminate them.
arXiv preprint arXiv:2411.11114 , year=
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 4roles
background 2polarities
background 2representative citing papers
Causal mediation analysis shows harmful LLM outputs arise in late layers from MLP failures and gating neurons, with early layers handling harm context detection and signal propagation.
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.
citing papers explorer
-
Why Do Aligned LLMs Remain Jailbreakable: Refusal-Escape Directions, Operator-Level Sources, and Safety-Utility Trade-off
Aligned LLMs exhibit Refusal-Escape Directions (RED) that enable refusal-to-answer transitions via input perturbations; these directions decompose exactly into operator-level sources, creating an inherent safety-utility trade-off when trying to eliminate them.
-
Why Do Large Language Models Generate Harmful Content?
Causal mediation analysis shows harmful LLM outputs arise in late layers from MLP failures and gating neurons, with early layers handling harm context detection and signal propagation.
-
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.
- SALLIE: Generation-Free Hidden-State Detection of Jailbreaks and Prompt Injections Across Text and Vision