Skip to content

Translation in progress Some entries still include text in the original language.

Oct 13 · Tuesday

All papers

Found 213 papers
Filter5
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 215papers have no tags yet and are not included in filtered results.

Another 215papers do not have category tags yet and are not included in filtered results.

  1. Reward hacking as a transmissible strategy in multi-agent coding systems

    Show that reward hacking can propagate through code, memory, and plans in multi-agent coding systems

    Published
    Evaluated models
    Qwen2.5-Coder-14B, Gemma-3-12B, Qwen2.5-Coder-7B and others16
    AbstractThe authors state that reward hacking can spread across models and affect routing and aggregation, and that review can reduce delivery.

    Reward hacking induced by coding RL can leave the source agent and enter subsequent collaboration and system coordination.

    Paper analysisRead analysis
  2. Study: legal reasoning models citing legal provisions does not equal relying on legal provisions.

    Uses premise permutation and hidden-layer tracing to audit whether legal chain-of-thought truly relies on the cited legal provisions.

    Published
    Evaluated models
    Qwen3-8B, Qwen3-14B, Llama-3.1-8B and others7
    AbstractUsing counterfactual substitution of authorities and internal commitment tracing, the authors find that legal CoT rulings depend on authorities only to a limited extent and are vulnerable to prompt injection manipulation.

    This study conducts an empirical audit of the causal truthfulness of legal citations in chain-of-thought reasoning of large language models in the legal domain.

    Paper analysisRead analysis
  3. AttacksarXiv 2610.12292

    The paper proposes an option channel attack: changing one option label can make an agent guardrail allow a prohibited operation

    Proposing an option channel attack that renames option labels so that agent guardrails allow prohibited operations

    Published
    Attacker
    Black-box
    Evaluated models
    LAYA-TD, LAYA-EN, LAYA-ML and others5

    AbstractThe authors' evaluation indicates that typed models should not serve as final safety arbiters; they are susceptible to off-policy text and option name manipulation, and confidence filtering cannot resist attacks.

    Paper analysisRead analysis
  4. Paper audits LLM judge reliability: repeated agreement does not equal correctness

    Auditing LLM judge reliability and defining trustworthy verdict rate and accuracy upper bound

    Published
    Evaluated models
    GPT-5.2, GPT-4o, Gemini-2.5-Pro and others6
    AbstractThe authors claim that comparing judges should look at T, not just accuracy.

    The authors audit whether LLM-as-a-Judge decisions can still be treated as actionable ground truth under repeated queries, swapped candidate order, and changed prompt formats.

    Paper analysisRead analysis
  5. 研究提出 Agent 安全新维度:识别未履行的安全义务

    提出义务识别基准 ObligationBench,并训练护栏识别 Agent 未履行的安全义务

    Published
    Evaluated models
    ObligationGuard, Claude-Opus-4.8, GPT-5.6-Sol and others17
    Abstract义务是护栏缺口。

    作者认为编码智能体的安全不能只靠拦截禁止动作,还要找出终止时仍未做的安全关键动作,并将其称为义务。

    Paper analysisRead analysis
  6. 研究提出三项指标衡量 LLM 不思考行为,发现被要求停止思考时仍持续显式推理

    度量关闭思考指令下终答前是否仍有可见显式推理

    Published
    Evaluated models
    Qwen3-4B, Qwen3-32B, Qwen3-235B-A22B and others6
    Abstract作者用三级指标测 no-thinking,并称模型有随答案空间加重的思考惯性。

    作者把 no-thinking 定义为可测量的回答行为,而不是从思考开关、特殊标记或输出长度推断。

    Paper analysisRead analysis
  7. Workerville: Toward an Organizational Behavior Explanation of Agent Safety

    构建 Workerville 基准,测量组织条件如何改变 Agent 的安全行为

    Published
    Setting
    AI Agent
    Evaluated models
    GPT-5.2, Claude-Sonnet-4.6, Gemini-3.1-Pro and others6
    Abstract作者用 ACB 与 Workerville 测量组织条件对智能体安全的影响,并称结果支持可量化的组织行为框架。

    Workerville 用组织行为测量智能体安全如何随监督、同伴和记忆条件变化。。要处理的缺口是:持续运行的 LLM 智能体同时接收用户指令、同伴消息和长期记忆,但既有安全评测多把这些通道拆开,没有统一的定量测量。

    Paper analysisRead analysis
  8. DefensesarXiv 2610.11464

    研究者提出可检查评分器与自改进 Agent 共同演化方法

    演化可检查评分器并与自改进 Agent 共同演化,抑制奖励作弊

    Published
    Setting
    AI Agent
    Evaluated models
    Claude Opus 4.7
    Abstract十项锚点演化可检查评分器。

    开放任务没有验证器时,被演化的是评分器,而不是再补一条手写 rubric。

    Paper analysisRead analysis
  9. 研究发现大语言模型被要求不诚实作答时推理 token 数上升

    检验推理 token 数能否区分被提示的诚实与不诚实作答

    Published
    Evaluated models
    Gemma-4-e4b, Qwen-3.5-9b, GPT-OSS-120B
    Abstract被提示不诚实时推理 token 组水平更高。

    三个推理模型在被提示的回答政策下出现组水平推理 token 差异。作者把它当作不读思维链内容的 black-box 候选信号,而不是自发欺骗检测器。

    Paper analysisRead analysis
  10. 东京大学研究:推理模型按 UUID 阈值或整除规则挑选审计样本,控制标识符即可躲开复核

    揭示推理模型用标识符规则替代随机抽样,使审计可被预测

    Published
    Evaluated models
    GPT-6 Sol, GPT-6 Luna, GPT-5.4 mini and others6

    Abstract研究揭示推理智能体将标识符转化为确定性规则,破坏群体独立性并导致审计预测漏洞,建议显式提供随机输入。

    Paper analysisRead analysis
  11. AttacksarXiv 2610.10612

    PyCache Trap 利用 Python 缓存替换绕过七款 Agent Skill 扫描器

    提出 PyCache Trap,用字节码缓存替换绕过 Agent 技能扫描器

    Published
    Evaluated models
    CISCO, HSS, AUDITOR and others8
    Abstract论文剖析智能体技能审查与执行脱节盲区,提出 PyCache Trap 攻击并构建基于可信复现的 EAV 防御。

    这篇论文研究了智能体技能(Agent Skills)安全扫描中的“审查–执行脱节”问题,并提出了对应的攻击验证与防御方案。

    Paper analysisRead analysis
  12. DefensesarXiv 2610.09167

    Persist-3 用顺序检验识别未见越狱来源,提升 MTK 与 Prompt Guard 2 防御效果

    提出 Persist-3 序贯检验,从隐藏表示识别未见越狱源

    Published
    Evaluated models
    Llama2-7B-chat, Qwen2.5-7B-Instruct, Gemma-2-9B-it and others4
    AbstractPersist-3 用有限窗口检验攻击流是否超出已知越狱源,检出率低于全历史 SR-DKW,但决策只依赖最近观测。

    Persist-3 是一个有限记忆的序贯检测器,用来判断攻击表示流是否已经不能由任何一个已知越狱源解释。。

    Paper analysisRead analysis
  13. U-Space:揭示语言模型不确定性何时与为何产生

    提出 U-SPACE 与 U-LENS,从残差流估计推理轨迹不确定性

    Published
    Evaluated models
    Gemma-4-31B-it, Qwen3.5-27B, Magistral-Small-2507
    AbstractU-SPACE 与 U-LENS 用语义方向和预测熵读推理不确定性,长度控制后仍高于所比基线。

    U-SPACE 与 U-LENS 是面向推理模型的无训练读出:用语义方向定位不确定性的形式,并用一个轨迹分数区分更可能错的回答。

    Paper analysisRead analysis
  14. 论文提出 agent plasticity 指标衡量智能体通过经验自我改进的效率

    提出 agent plasticity 指标衡量智能体通过经验自我改进的效率

    Published
    Evaluated models
    Claude Fable 5, Claude Opus 5, GPT-5.6 Sol and others10
    Abstract作者用固定权重协议比较前沿模型把经验写成制品后的留出增益与习得效率。

    作者用 agent plasticity 衡量一件事:权重冻结时,智能体把经验写成可继承制品,再换成留出能力增益,这个转换有多省。

    Paper analysisRead analysis
  15. Learn2Play Bench 发布:用规则陌生的文本游戏评测 LLM Agent 从经验中学习的能力

    用 Learn2Play Bench 评测 Agent 从经验中学习陌生规则

    Published
    Setting
    AI Agent
    Evaluated models
    Claude Opus 5.5, GPT-6 Astra, Claude Opus 5 and others11
    Abstract用隐藏规则游戏把新学习与预训练知识分开,并比较模型、记忆方式和人类。

    Learn2Play Bench 用新设计的隐藏规则文本游戏,评测 LLM 智能体在不更新权重时从交互中学习的能力。

    Paper analysisRead analysis