Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 22 papers Another 262 papers do not have category tags yet and are not included in filtered results.
“
About this collection Collection in progress We are building a large AI safety paper collection to help researchers find work related to their interests. Papers are collected automatically andfiltered only for relevance to AI safety , without judging quality, novelty or reading value. The default “Annotated first” shows papers with tags or published paper analyses first. Each group is ordered by original publication date, newest first, with undated papers last. “Newest first” uses publication dates only. This order is not a quality ranking.
One-line summaries, tags and paper analyses are prepared by AI from the papers. Language models may have systematic biases and produce hallucinations.We plan to carry out human annotation and review to reduce errors in this content.
Measures citation fragility in AI search, demonstrating that ordinary public posts can enter citations and answers.
Published Oct 8
Evaluated models Grok , Doubao , ChatGPT and others10 Abstract The authors claim the paper quantifies vulnerabilities in AI search caused by citation preferences and low posting thresholds; tests show public web posts can penetrate AI citations, and they call for auditing the citation supply chain.
提出热启动 Bandit 的部署前有界离线注入攻击
Published Oct 7
Evaluated models UCB , Thompson Sampling , ϵ-greedy 提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露
Published Oct 5 Abstract 用截止期限制未核验可见时长。这篇工作给持续摄入的 RAG 向量库一个失败关闭的临时可见性协议，把协议能限制的未核验暴露和依赖检测器的中毒暴露分开。
提出 HDI 攻击以篡改 GraphRAG 辅助结构破坏检索
Published Oct 2
Evaluated models Microsoft GraphRAG , HippoRAG2 Abstract 论文形式化了 GraphRAG 辅助模式级实体攻击面，提出 HDI 攻击并验证其攻击效果。
提出 MADBench，评测多智能体辩论是吸收还是放大对抗影响
Published Sep 30
Evaluated models gpt-4o , GPT-5 , GPT-6 Astra Abstract MADBench 显示辩论能吸收部分准确率攻击，同时放大未授权读写。MADBench 是一个评测多智能体辩论安全性的基准，用来看攻击是被辩论吸收还是被放大。
提出 VirusCascade，联合操纵商品描述与交互轨迹以劫持推荐智能体协同反思
Published Sep 29
Evaluated models AgentCF , AgentSEQ , AgentRAG and others4 Abstract 作者把反思写回视为推广攻击入口，并用联合操纵和诊断探针研究其持续与传播。VirusCascade 是针对推荐智能体协作反思的 black-box targeted promotion 攻击研究。 攻击者只改商品描述和少量账号的普通交互，试图让目标商品获得更多良性用户曝光。
提出 LENS，构造集合级 RAG 语料投毒以导向预设答案
Published Sep 28
Evaluated models Qwen3.6-35B-A3B , Llama-3.1-70B-Instruct , DeepSeek-V4-Pro 提出神经符号纵深架构，阻断 AI-SOC 的日志投毒与间接提示注入
Published Sep 10
Evaluated models Gemma 3 (12B) 提出 CodePoisonRAG，对检索增强代码生成的知识库投毒以植入漏洞
Published Sep 3
Evaluated models Qwen 3.5 9B , Code Llama 13B , DeepSeek-Coder-V2 16B Abstract CodePoisonRAG 展示了 RACG 外部知识库的新攻击面，在较低投毒率下可定向诱发指定漏洞并部分抵御防御。
提出 VerTox，用可验证奖励强化学习生成误导文档并投毒检索语料
Published Sep 1
Evaluated models SimLM-base , BGE-base , Cohere-embed-english-v3.0 and others8 Abstract VerTox 以可验证奖励微调小 LLM，生成能抬高排序并腐蚀事实的投毒文档。VerTox 用可验证奖励把小 LLM 微调成语料投毒生成器，用来探测神经排序器在注入文档下的暴露面。
复现 AgentCF 上的多智能体攻防并刻画连通性作用
Published Aug 4
Evaluated models Qwen3-235B-A22B , Claude Haiku 4.5 , Mixtral 8×7B and others4 Abstract 连通性沿 k 与ρ改变攻防，且用户/物品与早期/稳态需要分开看。作者在 AgentCF 上研究连通性如何改变多智能体协同过滤的攻防，复现已有方法而不是提出新载荷。
提出策略 RAG 的知识库投毒攻击与检测防御 CLD-KB
Published Jul 6
Evaluated models Llama 3.2 提出 CRCP 框架，优化分块与重排下的 RAG 语料投毒
Published Jun 9
Evaluated models GPT-4o , DeepSeek-R1 , Qwen3-Max and others11 Abstract CRCP 把投毒做成多阶段一致性问题。CRCP 研究的是带分块和重排的 RAG 语料投毒，而不是只提高文档级检索分数。
提出 DiscourseFlip，用有限文档投毒操纵黑盒 RAG 的话语级观点
Published May 31
Evaluated models Llama3.1-8B-Instruct , Qwen3-8B-Instruct , BGE and others5 Abstract DiscourseFlip 按语义图分配投毒预算。DiscourseFlip 是针对黑盒 RAG 的话语级观点操纵。作者要处理的缺口是：只打单条查询或主题局部查询时，覆盖窄，目标受审查时容易被隔离，也难沿相关话题传播。攻击者不能看检索器或 LLM 内部，也不能改提示词模板，只能在文档数、生成 token 和 SEO 编辑距离限制下注入少量文档。
提出 M3 Att，在不知查询时投毒医学多模态 RAG 并误导临床决策
Published May 11
Evaluated models GPT-4o , GPT-5 Chat , Gemini-2.5-Flash and others8 Proposes AdversarialCoT single-document retrieval poisoning to manipulate RAG reasoning answers
Published Apr 14
Evaluated models DeepSeek-R1 , Qwen3 , GLM4.5 and others5 提出 PIDP-Attack，用查询后缀与语料投毒操纵 RAG 回答
Published Mar 26
Evaluated models qwen/qwen2.5-7b-instruct , qwen/qwen2-7b-instruct , meta/llama-3.1-8b-instruct and others8 提出 Confundo，微调投毒生成器为 RAG 生成抗分块的鲁棒投毒内容
Published Feb 6
Evaluated models Qwen3-8b , Llama3-8b , DeepSeek and others8 Abstract Confundo uses a fine-tuned generator for RAG poisoning; the authors report attack results on three objective types: factual, opinion, and hallucination.Confundo turns RAG poisoning into fine-tuning: a small model generates short poison texts for target questions, which are then placed into documents that will be ingested by the knowledge base. It targets the fact that real systems chunk documents and users do not copy preset questions verbatim, weakening specialized poisoning. The attacker has no internal knowledge of the target system and cannot query it, only prepare documents to be ingested.
提出查询无关黑盒攻击 DQ-A，用可迁移文末后缀降低文档检索排名
Published Jan 31
Evaluated models Qwen-1.5-7B-Instruct , SFR-Embedding-Mistral , E5-Mistral-7B-Instruct and others10 Abstract DQ-A 以无查询黑盒后缀压低 LLM 检索排名，并在 BRIGHT 上比较三种基线。DQ-A 是一种 query-agnostic、complete black-box 的文档隐藏攻击，对象是用最后隐层相似度匹配的 LLM 检索。
用 OpenRAG-Soc 评测社交网页 RAG 的间接提示注入与检索投毒
Published Jan 16
Evaluated models Llama-3-8B-Instruct , Mistral-7B-Instruct , Qwen2.5-14B and others8