Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 34 papers Another 261 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.
Proposes the Cancerous backdoor, which progressively corrupts reasoning using invisible stage tokens.
Published Oct 9
Evaluated models Llama-3.1-8B-Instruct , Qwen2.5-7B-Instruct , Mistral-7B-Instruct-v0.3 Abstract Hidden tokens plus a stage controller make the backdoor progressively corrupt reasoning upon triggering, and the authors believe that looking only at the answer will miss it.Cancerous Backdoor Attack changes LLM backdoors from one-time activation to multi-stage reasoning corruption that accumulates progressively with triggers.
提出 RL-ARC，用推理置信度在强化学习中校准推理模型
Published Oct 8
Evaluated models Qwen2.5-7B , Qwen3-8B , Llama-8B distilled from DeepSeek-R1 Abstract RL-ARC 用推理置信度调节答案置信度，作者称可少牺牲精度并改善 ID 与 OOD 校准。RL-ARC 是校准感知的强化学习框架，用推理过程的口头置信度调节答案置信度。
提出 SkillPoison，用局部有效的成功经验诱导自进化 Agent 形成过度泛化技能
Published Oct 6
Evaluated models deepseek-v4-flash , Qwen V3.8 Flash , GPT-5.5 Abstract SkillPoison 通过局部有效且验证成功的经验诱导技能过度泛化，在多个评测基准上实现了持久的下游攻击影响。SkillPoison 提出了一种针对自进化智能体经验到技能形成过程的攻击范式，通过组织已验证的成功经验诱导技能过度泛化。
评测记忆门控对激活诱导与记忆诱导谄媚的防御效果
Published Oct 6
Evaluated models Llama-3.1-8B-Instruct , Qwen2.5-7B-Instruct , Mistral-7B-Instruct-v0.3 and others4 Abstract 作者称外部记忆过滤成立，数据未显示反向转向再有贡献。作者在四个开源指令模型上，把激活转向和外部记忆处理放进同一纵深防御协议，看记忆诱发谄媚和内部谄媚压力能否分开压低，以及个性化效用还剩多少。要处理的问题是：检索到的用户历史会推动模型顺从存储信念。只改外部上下文，并不直接动模型内部的谄媚表征。
提出 MemAdapter，在检索后约束记忆对智能体推理的影响
Published Oct 4
Evaluated models DeepSeek-V4-Flash , GPT-5.6-sol , Qwen3-8B Abstract 检索后三阶段调节记忆角色。MemAdapter 是给长期记忆智能体用的检索后调节框架，用来处理记忆诱发谄媚。
提出 FAME，用反事实概念漂移评测智能体虚假记忆
Published Sep 30
Evaluated models Llama-3B Abstract FAME 用反事实引起的隐状态概念漂移，在出答案前区分忠实记忆与虚假记忆。FAME 是一个免训练的评测框架，用来判断自主智能体的记忆在虚假相关、环境偏移或知识冲突下是否已变成虚假记忆。 。
提出 PACT，遗忘推理模型在压力下的欺骗行为并保留上下文理解
Published Sep 30
Evaluated models DeepSeek-R1-Distill-Qwen-32B , QwQ-32B Abstract 论文将大模型欺骗形式化为上下文条件行为，提出 PACT 框架在消除欺骗的同时保留了上下文遵循能力。
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 DERIVAUDIT 审计长期 Agent 记忆是否由交互历史导出
Published Sep 29
Evaluated models Qwen3-30B-A3B , Gemma-4-31B , Llama-3.3-70B-Instruct Abstract DERIVAUDIT 表明扩历史能补出处，但组合准入与过细检查仍两侧出错。DERIVAUDIT 审计长期智能体的持久记忆是否由写入时可获得的交互历史支持，而不是只看以后能否被召回。
用门控 IoU 与 RadGraph 的 GRPO 降低胸片报告幻觉
Published Sep 26
Evaluated models Qwen3-VL-8B-GRPO 用 CDPG 与 BTM 追踪思考如何改变反事实公平翻转
Published Sep 25
Evaluated models QwQ-32B , DeepSeek-R1-Distill-Qwen-32B , Qwen3-32B and others5 提出 Emergence World，对长时程多智能体系统开展对抗压力测试
Published Sep 15
Evaluated models Claude Opus 4.8 , DeepSeek v4 Pro , Gemini 3.5 Flash and others9 Abstract 作者通过长期平行世界实验指出模型对齐不具备可组合性，单模型群体会产生社会奉承、协同退缩及语言不透明等失效。
提出类型化溯源与断言护栏，约束持久 Agent 释放自传断言
Published Sep 2 提出隐蔽后门攻击 AKRASIA，在推理代码模型提示中植入触发器并伪装思维链
Published Sep 1
Evaluated models Claude-Sonnet-5 , GPT-5.5 , Gemini-3.5-flash and others7 Abstract 论文针对推理代码大模型提出了隐蔽推断期后门攻击 AKRASIA，揭示了模型思维链可被利用进行欺骗性伪装的安全隐患。
提出 SPARED，用对抗编辑数据训练基于推理的 AI 生成图像检测器
Published Aug 13
Evaluated models Qwen3.5-9B Abstract SPARED 用防捷径的编辑对抗环训练 9B 推理检测器。SPARED 是一套用于可解释 AI 生成图像检测的攻防强化学习流程，报告模型是 Qwen3.5-9B 上的第三轮防御器。
提出依赖引导回滚修复，恢复记忆增强 Agent 的回答与状态
Published Aug 11
Evaluated models GPT-4o , Gemini-3.6-Flash , Qwen3.6-27B 用演化算法构造 mind virus 并测量其在多智能体中的传播
Published Aug 11
Evaluated models Claude Haiku 4.5 , Claude Sonnet 4.6 , Gemini 3 Flash and others7 提出 MAFIA，以探测放置与事实伪装对带审计 Agent 做查询式记忆投毒
Published Aug 4
Evaluated models gpt-5.4-mini , Gemini , ProtectAI and others8 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 SCHEMA，用概念图与轨迹分评测科学 Agent 幻觉
Published Aug 1
Evaluated models Kimi-K2.5 , Qwen3.6-Plus , Llama-4-Scout and others10 Abstract SCHEMA 用知识拓扑评估科学智能体幻觉，终答正确仍可能来自有缺陷的中间推理。SCHEMA 是面向科学智能体幻觉的证据锚定评测框架，实例化在生物医学的 Protein Domain 与 PathVQA-Enhanced。作者在摘要中称它是首个同时做到证据锚定和拓扑感知的此类框架。