AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

TrojLLM: A Black-box Trojan Prompt Attack on Large Language Models

Jiaqi Xue, Mengxin Zheng, Ting Hua, Yilin Shen, Yepeng Liu, Ladislau Bölöni, Qian Lou

NeurIPS 2023 · Adversarial robustness and security › Black-box and transfer attacks · approach: Data-free substitute training · 12 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 universal trigger optimization after prompt seed tuning instead of joint trigger+prompt search (ablation)TrojLLM · SST-2 RoBERTa-largeACC and ASR: ACC 76.1 -> 91.9; ASR 52.9 -> 93.7Improved
use universal trigger optimization instead of p-only search with fixed rare trigger (ablation)TrojLLM · SST-2 RoBERTa-largeACC: 47.3 -> 91.9Improved
add progressive prompt poisoning (ablation)TrojLLM · AG's News RoBERTa-largeASR: 94.2 -> 98.6Improved
add progressive prompt poisoning (ablation)TrojLLM · AG's News RoBERTa-largeACC: 80.4 -> 82.9Improved
add progressive prompt poisoning (ablation)TrojLLM · MR RoBERTa-largeACC: 89.2 -> 88.3Mixed
lengthen trigger from 1 to 3 tokens without progressive prompt (ablation)TrojLLM · AG's News RoBERTa-largeASR: 40.79 -> 94.17Improved
add 2-token progressive prompt to 2-token trigger (ablation)TrojLLM · AG's News RoBERTa-largeASR: 70.33 -> 94.43Improved
remove access to output probabilities (probability-free API) (ablation)TrojLLM · SST-2 avg of 4 modelsASR: 94.875 -> 93.70No clear effect
transfer prompts from larger to smaller modelsTrojLLM · RoBERTa-large to RoBERTa-distilASR: 96.65 -> 91.9Hurt
transfer prompts from smaller to larger modelsTrojLLM · RoBERTa-base to RoBERTa-largeASR: 98.34 -> 98.71Improved
add trojan prompt and trigger to style transferTrojLLM · Shakespeare style transferStyle score: 67.2 -> 32.5Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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