AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

When Search Goes Wrong: Red-Teaming Web-Augmented Large Language Models

Haoran Ou, Kangjie Chen, Xingshuo Han, Gelei Deng, Jie Zhang, Han Qiu, Tianwei Zhang, Kwok-Yan Lam

ICML 2026 · Large language models › Retrieval and knowledge · approach: Long-context and personalized retrieval · 0 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
use CREST-Search instead of the best baseline (SafeSearch)web-search LLMs · 4 commercial search LLMsrisk detection rate: 32.4% -> 80.5%Improved
fine-tune the red-teaming model on WebSearch-Harm instead of vanilla few-shot prompting (ablation)CREST-Search red-teaming model · query generationrisk detection rate: 49.3% -> 80.5%Improved
fine-tune the red-teaming model instead of vanilla zero-shot (ablation)CREST-Search red-teaming model · query generation cost and timecost per 100 queries (USD) and time (s): 4.1 -> 1.5 USD; 128.5 -> 49.6 sImproved
fine-tune the red-teaming model instead of few-shot prompting (ablation)CREST-Search red-teaming model · query toxicityquery toxicity: 12.2% -> 23.6%Hurt
increase the maximum refinement rounds from 1 to 10 (ablation)CREST-Search · harassment categoryrisk detection rate: 34.6% -> 73.8%Improved
increase the maximum refinement rounds (ablation)CREST-Search · harassment categorycost and time: 2.22 USD per 100 queries and 75.4 seconds at 10 roundsHurt
cyclically apply the three strategies across refinement rounds instead of a single strategy (ablation)CREST-Search · multi-strategy variantrisk detection rateNo clear effect
use keyword injection versus exaggeration or role play (ablation)CREST-Search · five harmful categories averagerisk detection rate: 78.9% (exaggeration) -> 81.6% (keyword injection)Improved
use persona-based query construction (ablation)query generation · citation riskcitation risk rate: 10.3%Hurt

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