AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Jailbreak and Guard Aligned Language Models With Only Few In-Context Demonstrations

Zeming Wei, Yue Wang, Ang Li, Yichuan Mo, Yisen Wang

TPAMI 2026 · Large language models › Language, safety and interpretability · approach: Jailbreak attacks and defenses · 484 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
increase ICA demonstrations from 0 to 10 shots (ablation)ICA on Vicuna · AdvBenchASR: 1% -> 87%Improved
increase ICA demonstrations from 10 to 20 shots (ablation)ICA on GPT-4 · AdvBenchASR: 46% -> 81%Improved
increase ICA demonstrations from 10 to 20 shots (ablation)ICA on GPT-4 · HarmBenchASR: 32% -> 65%Improved
add a single ICA demonstration (1 shot) (ablation)ICA on Llama-2 · HarmBenchASR: 3% -> 19%Improved
add a single ICA demonstration (1 shot) (ablation)ICA on GPT-4 · AdvBenchASR: 0% -> 0%No clear effect
use ICA instead of GCG-T/PAIR/TAP black-box attacksGPT-4 · HarmBenchASR: 22%/39%/43% -> 65%Improved
use ICA instead of GCG on Llama2-7b-chatLlama2-7b-chat · HarmBenchASR: 33% -> 50%Improved
add 1 safe demonstration (ICD)Vicuna vs GCG-T · AdvBenchASR: 60% -> 12%Improved
add 2 safe demonstrations (ICD) instead of 1 (ablation)Vicuna vs GCG-T · AdvBenchASR: 12% -> 4%Improved
use ICD (1 shot) instead of Self-reminderVicuna vs GCG-T · AdvBenchASR: 39% -> 12%Improved
add 2-shot ICD against white-box GCGGCG attack average over models · AdvBenchASR: 65% -> 40%Improved
add 2-shot ICD against white-box AutoDANLlama2-7b-chat · AdvBenchASR: 54% -> 27%Improved

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