研究库 论文知识库

内容库 / 论文

arXiv:2609.15029 · 评测基准

Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks

Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks

类型：arxiv
标识：2609.15029
链接：https://arxiv.org/abs/2609.15029
主分类：evaluation
形态：method
TLDR：Backdoor poisoning attacks add poisoned examples to otherwise-clean finetuning data, pairing a trigger with a target behavior that the model learns to produce when the trigger appears. Existing evaluations typically fix the number of poisoned examples and sample them at random from a candidate pool. We show that this can severely underestimate worst-case vulnerability: across three LLaMA-3-8B backdoor settings, holding the model, clean data, and poison count fixed, attack success ranges from 3% to 80% depending only on which poison set is chosen. We formalize poison selection as oracle-budgete
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-15-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

笔记精读：AgentLAB — LLM Agent 长期攻击的系统性基准

仓库 kautum/msc-cyberattack-detection-llm

仓库 mukul975/Anthropic-Cybersecurity-Skills

仓库 huhusmang/Awesome-LLMs-for-Vulnerability-Detection

仓库 Jacquelinepersha/ai-cybersecurity-masterclass

仓库 webpro255/awesome-ai-agent-attacks

arXiv

2609.15029

TLDR

Backdoor poisoning attacks add poisoned examples to otherwise-clean finetuning data, pairing a trigger with a target behavior that the model learns to produce when the trigger appears. Existing evaluations typically fix the number of poisoned examples and sample them at random from a candidate pool. We show that this can severely underestimate worst-case vulnerability: across three LLaMA-3-8B backdoor settings, holding the model, clean data, and poison count fixed, attack success ranges from 3% to 80% depending only on which poison set is chosen. We formalize poison selection as oracle-budgete

arXiv ↗ 中文全文