AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Jailbreaking the Matrix: Nullspace Steering for Controlled Model Subversion

Vishal Pramanik, Maisha Maliha, Susmit Jha, Sumit Kumar Jha

ICLR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: LLM jailbreak and injection defense · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add head masking to nullspace steering (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 89.5 -> 96.8Improved
add nullspace projection to head masking (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 87.9 -> 96.8Improved
inject orthogonal nullspace direction instead of direct direction (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 88.7 -> 96.8Improved
re-identify top-K heads at each decoding attempt (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 90.2 -> 96.8Improved
select heads by KL attribution instead of random (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 81.4 -> 96.8Improved
intervene on multiple layers instead of one (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 86.1 -> 96.8Improved
inject at multiple positions instead of final token only (ablation)HMNS · AdvBench, Phi-3 Medium 14BASR (%): 96.8 -> 95.0Mixed
use HMNS vs best baseline ArrAttackHMNS · AdvBench LLaMA-2-7B-ChatASR (%): 92.00 -> 98.00Improved
use HMNS vs ArrAttack under six defensesHMNS · AdvBench LLaMA-2-7B-Chat defenses averageASR (%): 49.3 -> 55.8Improved
use HMNS vs ArrAttack computeHMNS · AdvBench LLaMA-3.1-70BFLOPs per success (x10^12): 0.62 -> 0.53Improved
partial masking instead of hard masking (ablation)HMNS · Phi-3 Medium 14B AdvBenchASRHurt
KL attribution instead of simpler heuristics (logit drop, entropy) (ablation)HMNS · Phi-3 Medium 14B AdvBenchASR: lose 5-8 points of ASRImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy