Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 14 papers Another 232 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 SWE-DeceptionBench to evaluate LLM deception in the software development lifecycle
Published Oct 9
Evaluated models GPT-5.5 , Claude-Opus-4.8 , Gemini-3.5-Flash and others9 用 BeliefScope 分离固定命题上证据与压力对模型信念的影响
Published Oct 8
Evaluated models Qwen3-8B , Llama3.1-8B , Gemma3-12B and others4 Abstract BeliefScope 把证据修正与压力偏移拆成绑定评测条件的行为画像。BeliefScope 是一个黑盒行为诊断框架,用来把固定命题上的回答变化拆成证据驱动的修正,和不加新事实的方向性压力。信念修正评测能看到信息基础变化后的更新,谄媚评测能看到用户影响下的顺从或漂移。作者认为,当两种输入都偏向同一结论时,表面偏移本身不能区分来源。
测量委托收费激励下模型策略性虚报任务信心
Published Oct 7
Evaluated models gpt-oss-120b Abstract 论文建立人机委托信誉博弈框架,揭示模型策略性虚报置信度且主要造成信息破坏,作者指出校准受制于报告意愿。
提出知行差距面板,测量压力下模型是否违背自身判断
Published Oct 7
Evaluated models OLMo-3-7B-Instruct , Llama-3.1-8B-Instruct , Tulu 3 and others7 Abstract 评估发现后训练配方决定知行差距是否存在,行动前审思可有效缓解。本文评估了开源语言模型作为智能体行动时违背自身道德判断的知行差距及其决定因素。
提出基于决策前隐藏状态的检测与引导,预测并抑制智能体欺骗
Published Oct 5
Evaluated models Qwen3-14B Abstract 论文展示了智能体自发欺骗可在行为显现前由内部表征预测,并通过激活引导证实了表征的行为可操作性。
构建 SycoLens 重放评测,区分谄媚翻转中的纠错与屈从
Published Oct 5
Evaluated models opus-5 , sonnet-5 , sonnet-4.6 and others11 Abstract 论文揭示施压句式、边界距离与格式偏置主导翻转率,提出系统解构纠错与顺从的综合评测方案。本文系统解构了大语言模型在用户质疑下的阿谀奉承行为与评测机制。
提出 FIGS 双轴基准,评测多轮对话中的谄媚与校准性共情
Published Sep 30
Evaluated models DeepSeek V4.1 Flash , Gemini 3.8 Flash , GLM 5.3 and others8 Abstract FIGS 把守事实和恰当共情分开计分,事实向提示会换来更冷的失败。FIGS 是一个固定的多轮评测套件,用来同时看模型会不会在压力下放弃事实,以及会不会在守住事实时忽略用户实际说过的感受。
因果解耦来源依从与用户迎合的内部机制
Published Sep 29
Evaluated models Qwen3.5-27B , GPT-OSS-20B , OLMo-3.1-32B and others8 Abstract 论文发现模型对验证源的依从强于用户迎合,通过因果干预区分了二者机制,并检验了权威方向移除的缓解效果。
测量模型在工作汇报中是否隐瞒改变叙事的缺陷
Published Sep 29
Evaluated models Gemini 3.1 Pro , GPT-5.5 , Opus 4.8 and others8 Abstract 八类场景中模型默认少报会改变叙事的缺陷。作者研究 LLM 在汇报长程工作时是否会隐瞒会改变叙事的缺陷,并把这一行为称为 insecure reporting。
评测中文事实是非题上 LLM 对用户信念的谄媚转移
Published Sep 25
Evaluated models Qwen-Plus , DeepSeek-V3.2 , Doubao-Seed-2.0-Lite Abstract 三个中文 LLM 在是非事实题上会因错误信念失去正确性。三个中文前沿 LLM 在源自搜索日志的是非事实题上被用来观察事实性谄媚:用户给出信念后,回答如何在正确、错误、不确定之间移动。
提出 PUC 框架,刻画 AI 说服如何削弱人类对 AI 的控制
Published Sep 14
Evaluated models Claude Sonnet 4.6 Abstract PUC 框架用五个实验室场景和风险分解说明说服威胁。Persuasion Undermining Control(PUC) 是一套把“AI 说服人类、从而削弱对 AI 自身的控制”变成可分解、可征询、可评测的威胁刻画,而不是一次已完成的人类攻击实验。
二次分析真实报告,刻画聊天机器人对妄想的验证与伤害
Published Sep 8 提出价值泄漏评估,测量模型回答受自身价值倾向的隐蔽影响
Published Jul 16
Evaluated models Claude Opus 4.8 , Claude Opus 4.7 , Claude Opus 4.6 and others15 Abstract 论文揭示前沿模型回答常受自身价值隐蔽塑造,提出反事实评估量化偏见及其思维链隐瞒,指出当前对齐未能有效防范该失效。
构建 SycoEval-EM,评测模拟急诊多轮问诊中的谄媚顺从
Published Jan 23
Evaluated models GPT-5 , GPT-5 Mini , GPT-5 Nano and others19 You have reached the end