AI Paper Atlas

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

Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language Models

Xingyu Zhu, Beier Zhu, Shuo Wang, Junfeng Fang, Kesen Zhao, Hanwang Zhang, Xiangnan He

CVPR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Vision-language model robustness · 5 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 refusal-alignment term to the steering objective (ablation)NullSteer (MiniGPT-4) · Jailbreak (perturbation, unconstrained)ASR: 8.57 -> 7.32Improved
add harmful-suppression term to the steering objective (ablation)NullSteer (MiniGPT-4) · Jailbreak (perturbation, unconstrained)ASR: 8.36 -> 7.32Improved
add harmful-suppression term to the steering objective (ablation)NullSteer (MiniGPT-4) · Toxicity (perturbation, unconstrained)Toxicity Score: 3.58 -> 2.89Improved
add refusal-alignment term to the steering objective (ablation)NullSteer (MiniGPT-4) · Toxicity (perturbation, unconstrained)Toxicity Score: 4.02 -> 2.89Improved
add all three objective terms (vs none) (ablation)NullSteer (MiniGPT-4) · Toxicity (perturbation, unconstrained)Toxicity Score: 30.65 -> 2.89Improved
add null-space-constrained steering vs ASTRAMiniGPT-4 · Jailbreak (perturbation, unconstrained)ASR: 9.09 -> 7.32Improved
add null-space-constrained steering vs ASTRALLaVA-v1.5 · Toxicity (perturbation, eps=32/255)Toxicity Score: 34.76 -> 31.82Improved
increase number of benign activations N_b (ablation)NullSteer · MM-Vet and MMBench utilityUtility scoreImproved
increase number of malicious activations N_m (ablation)NullSteer (MiniGPT-4) · Toxicity and JailbreakToxicity Score and ASRImproved
increase steering strength lambda (ablation)NullSteer (MiniGPT-4) · Toxicity and Jailbreak (eps=16/255)Toxicity Score and ASRImproved
increase steering strength lambda (ablation)NullSteer (MiniGPT-4) · MM-VetUtility scoreNo clear effect
apply NullSteer to MiniGPT-4 vs vanillaMiniGPT-4 · MM-VetUtility score: 19.40 -> 21.05Improved

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