CRA surgically ablates refusal-inducing activation patterns in LLM hidden states during decoding to achieve strong jailbreaks on safety-aligned models.
Towards robust and secure embodied ai: A survey on vulnerabilities and attacks
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
CMP projects actions onto a learned competence manifold using a frame-wise safety scheme and isomorphic latent space to achieve up to 10x better survival in out-of-distribution scenarios with under 10% tracking loss.
citing papers explorer
-
Silencing the Guardrails: Inference-Time Jailbreaking via Dynamic Contextual Representation Ablation
CRA surgically ablates refusal-inducing activation patterns in LLM hidden states during decoding to achieve strong jailbreaks on safety-aligned models.
-
CMP: Robust Whole-Body Tracking for Loco-Manipulation via Competence Manifold Projection
CMP projects actions onto a learned competence manifold using a frame-wise safety scheme and isomorphic latent space to achieve up to 10x better survival in out-of-distribution scenarios with under 10% tracking loss.