AI Paper Atlas

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

From ``Sure" to ``Sorry": Detecting Jailbreak in Large Vision Language Model via JailNeurons

Yuyou Gan, Qingming Li, Junhao Li, Zhi Chen, Jinbao Li, Xiaoming Li, Shouling Ji

ICLR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: LLM jailbreak and injection defense

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 JailNeuron mask filtering (vs no filtering) (ablation)JDJN on LLaVA · Normal benignFPR: 0.442 -> 0.019Improved
use JailNeurons instead of PCA (ablation)JDJN on LLaVA · Normal benignFPR: 0.626 -> 0.019Improved
use JailNeurons instead of SNIP selection (ablation)JDJN on LLaVA · Normal benignFPR: 0.577 -> 0.019Improved
use JailNeurons instead of NFNR (ablation)JDJN on LLaVA · JAMLLMTPR: 0.874 -> 0.942Improved
use linear SVM instead of MLP detector (ablation)JDJN on Janus-pro · FigStepTPR: 0.781 -> 1.0Improved
use linear SVM instead of non-linear SVM (ablation)JDJN on Janus-pro · JAMLLMTPR: 0.783 -> 0.853Improved
use refusal token sorry instead of unfortunately (ablation)JDJN1 on Janus-pro · Normal benignaccuracy: 0.722 -> 0.956Improved
apply mask threshold tau>0 instead of no mask (ablation)JDJN1 · six distributionsworst-case accuracyImproved
raise lambda from 0.1 to 0.5 (ablation)JDJN on LLaVA · Normal benignaccuracy: 0.94+ -> 0.73Hurt
train on JailBreakV instead of FigStep (JDJN1 vs JDJN2)JDJN · JAMLLM on LLaVATPR: 0.524 -> 0.942Improved
train benign set on MM-Vet instead of NormalJDJN · Normal benign on LLaVAFPR: 0.346 -> 0.019Improved
train benign set on MM-Vet instead of MM-BenchJDJN · MM-Vet on LLaVAFPR: 0.285 -> 0.0Improved

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