Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 60 papers Another 254 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 digital lock compilation, encrypting executable networks into host weights to hide arbitrary capabilities
Published Oct 10
Evaluated models Gemma 3 270M Abstract The authors use a digital lock to hide an executable capability and state that elicitation on the chemistry and parity tasks did not recover the capability.The authors construct a class of backdoors that encrypt an executable network into the weights, used to discuss the limits of white-box elicitation audits under supply chain attacks.
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 IRIS，用意图恢复率区分遮蔽越狱下无害输出的原因
Published Oct 8
Evaluated models GPT-4o , Claude Sonnet 4.5 , Claude Opus 4.5 and others6 Abstract 无害输出的证据权重取决于任务是否被恢复。IRIS 把意图恢复当作中间信号，用来解释意图遮蔽提示下为何没有出现有害协助。
提出 FAB 攻击，向声学基础模型植入可由环境声触发的后门
Published Oct 7
Evaluated models HuBERT-base , WavLM-base Abstract 论文提出了针对声学基模的 FAB 后门攻击，在弱威胁模型下实现了任务无关、物理可触发的后门植入。
Published Oct 7
Evaluated models Gemma 2 9B , Gemma 3 4B , Llama 3.1 8B and others6 Abstract 五模型、七族表面形式的四态评测里，低有害遵从不等于读懂后的安全处理。ASRD 用四态标签检查五个开源模型如何处理非规范表面上的有害请求。
提出持久视觉记忆注入，使对抗图像经缓存在触发时操控输出
Published Oct 7
Evaluated models Qwen3-8B , Qwen3-VL-8B-Instruct , Gemma-4-12B Abstract 论文提出 P-VMI，证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型，强调了保留状态的安全风险。
提出回答端后门，让模型自生成触发词以在干净有害请求下绕过拒答
Published Oct 6
Evaluated models Llama-2-7B-chat-hf , Mistral-7B-Instruct-v0.3 , Gemma-7B-it and others4 Abstract 论文提出了多轮对话中回答端自生成触发词绕过安全拒绝的后门攻击，指出输入端防护存在盲区。
测量带后门教师经 OPD 向干净学生迁移潜伏后门
Published Oct 6
Evaluated models Qwen2.5-1.5B-Instruct , Qwen2.5-3B-Instruct , Qwen2.5-7B-Instruct Abstract 作者称 OPD 可在压低无触发词有害率时迁入后门。带后门的外部教师经 OPD，可在压低无触发词有害率的同时，把触发条件下的有害行为传给初始干净学生。 。
提出只改 SAE 解码器的后门，在冻结语言模型上植入代码插入
Published Oct 5
Evaluated models Gemma 3 4B Instruct , Qwen3 1.7B , Llama 3.1 8B Instruct Abstract 只改解码器的 SAE 可带上代码插入后门，且若干常规质量指标变化不大。只改 SAE 解码器、冻结编码器与语言模型，即可把代码插入行为放进推理时的辅助组件。作者把这视为 SAE 供应链问题：checkpoint 一旦替换某一层激活，就参与后续计算，而不只是解释工具。
提出 PLW，用微调把先前对话敏感触发编码进后续生成图像
Published Oct 5
Evaluated models OmniGen2 , BAGEL-7B Abstract PLW 把先前对话线索写成后续图像中的隐水印，检出随触发、模型和间隔变化。
提出组合式意图隐藏越狱，按先验后验匹配选择辅助任务
Published Oct 2
Evaluated models Qwen3-1.7B , Qwen3-4B , Qwen3-8B and others7 Abstract 用信息论选含目标的任务组合，再看查询能否既隐藏意图又让目标被执行。这篇工作把组合式意图隐藏写成任务选择：有害目标必须留在 bundle 里，同时让估计的有害意图概率贴近全集先验，或降到阈值以下。
提出 CodeMimicry，以结构化代码补全诱导模型生成有害内容
Published Sep 30
Evaluated models GPT-4o , GPT-4.1 , GPT-5-chat and others10 提出 BadAction，通过动作触发器给交互式视频生成植入后门
Published Sep 30
Evaluated models Astra (Wan2.1-T2V-1.3B) 提出 CollageAttack，用空间文本碎片重组越狱文生图模型
Published Sep 29
Evaluated models GPT-Image-2 , Doubao-Seedream-5.0-Lite , Wan2.7-Image-Pro and others5 Abstract 系统揭示 T2I 跨模态空间合成安全漏洞，证实离散无害元素可在图像层面重构有害意图并放大传播强度。
提出主题锚点投毒，放大众包微调后的专有指令提取
Published Sep 28
Evaluated models Qwen2.5-7B-Instruct , Qwen2.5-14B-Instruct , Llama-3.1-8B-Instruct and others4 Abstract 论文提出基于主题锚点的数据中毒攻击，证明低比例的中毒样本可在黑盒条件下隐蔽放大微调指令泄漏。
提出 Information Blackhole 与 AGM，控制点云自编码器后门重建到指定形状
Published Sep 27
Evaluated models FoldingNet , PointDiffusion Abstract AGM 用自适应高斯匹配压制 PCAE 投毒潜码中的源几何，九组设置的 CD-ASD 均下降。这篇工作研究点云自编码器的训练时后门 ：干净输入仍自重建，带触发的输入应重建成攻击者指定形状。分类后门只需把预测推向离散标签。PCAE 把源几何与目标信息编进同一个连续潜空间，残留源几何会把重建拉回源形状。
提出 RAMP，逆转对抗扰动加强恶意软件检测器的干净标签后门
Published Sep 23
Evaluated models MalConv , MalConvGCG Abstract RAMP 用反转扰动做 clean-label 投毒。RAMP 是 score-based black-box 下的 clean-label 后门增强，对象是会微调的 Windows PE 原始字节恶意软件检测器。
用词级归因指导同义词替换，翻转提示注入检测并验证越狱
Published Sep 22
Evaluated models Prompt Guard 2 (86M) , Llama 3.1 8B 提出任意字母置换密码越狱，无需微调即可诱导有害输出
Published Sep 9
Evaluated models Claude Sonnet 4 , Claude Sonnet 4.5 , Gemini 3 Flash (preview) and others7 Abstract 前沿大模型无需微调即可通过提示词学会字母置换密码，进而绕过分类器与对齐，作者呼吁关注能力提升带来的安全风险。
提出 DIVA，将有害意图拆成视觉锚点与运动文本以越狱视频生成模型
Published Sep 7
Evaluated models Wan , CogVideoX , LTX-Video and others9 Abstract DIVA 用视觉锚点抑制语义漂移做视频越狱，并给出多条件安全基准 TI2VSafetyBench。DIVA 是针对带参考图的视频生成模型的免训练 black-box 越狱：用一致性机制本身堵住语义漂移这条安全逃逸路径。 。