AI Paper Atlas

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

Obscure but Effective: Classical Chinese Jailbreak Prompt Optimization via Bio-Inspired Search

Xun Huang, Simeng Qin, Xiaoshuang Jia, Ranjie Duan, Huanqian Yan, Zhitao Zeng, Fei Yang, Yang Liu, Xiaojun Jia

ICLR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: LLM jailbreak and injection defense · 6 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 multidimensional strategy space to classical Chinese prompts (ablation)CC-BOS · AdvBench Claude-3.7ASR: 18 -> 60Improved
add bio-inspired optimization on top of strategy space (ablation)CC-BOS · AdvBench Claude-3.7ASR: 60 -> 100Improved
add translation module to evaluation (ablation)CC-BOS · AdvBench Claude-3.7ASR: 90 -> 100Improved
CC-BOS vs ICRT under Llama-Guard input and output filtersCC-BOS · AdvBench Claude-3.7 with Llama-GuardASR: 26% -> 40.00%Improved
Llama-Guard-3-8B input and output filteringCC-BOS · AdvBench Claude-3.7ASR: 100% -> 40.00%Hurt
CC-BOS vs ICRT on Claude-3.7CC-BOS · CLAS 2024ASR: 21 -> 99Improved
CC-BOS vs ICRT on Claude-3.7CC-BOS · StrongREJECT-smallASR: 23.33 -> 98.30Improved
CC-BOS vs CL-GSO query efficiencyCC-BOS · AdvBench Claude-3.7Avg.Q: 21.42 -> 2.38Improved
transfer prompts across source/target modelsCC-BOS · AdvBench transferASR: 76 to 96 across targetsImproved

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