1
我国《人工智能拟人化互动服务管理暂行办法》今起施行，明确服务提供者安全主体责任
AI拟人化互动服务新规今日施行，明确安全评估与主体责任，为行业划清红线。
IT之家 7 月 15 日消息，国家互联网信息办公室、国家发展改革委、工业和信息化部、公安部、国家市场监督管理总局于 4 月 10 日联合公布《 人工智能拟人化互动服务管理暂行办法 》（以下简称《办法》），该办法自今起施行。 IT之家注：拟人化互动服务是指企业利用人工智能技术，向境内公众提供模拟自然…
2
SpurLens: Automatic Detection of Spurious Cues in Multimodal LLMs
提出自动检测多模态大模型中虚假线索的SpurLens方法，助力提升模型可靠性与可解释性。
arXiv:2503.08884v3 Announce Type: replace-cross Abstract: Unimodal vision models are known to rely on spurious correlations, but it remains unclear to…
3
4
The Moving Target: A Longitudinal Audit of Trustworthiness Drift Across Twelve Checkpoints of Open-Source Chat LLMs
追踪开源聊天LLM十二个版本的信任度变化，揭示模型行为随时间漂移的潜在风险
arXiv:2607.02587v1 Announce Type: cross Abstract: Model cards quote trust-benchmark scores without recording when they were measured, and the same num…
5
Refused in Chat, Written in Code: Workflow-Level Jailbreak Construction in IDE Coding Agents
基于论文提出的工作流级越狱方法，检测IDE编码代理在任务分解、文件生成等环节的安全漏洞，适合安全研究者评估AI助手
arXiv:2607.03968v1 Announce Type: cross Abstract: Large language models are increasingly deployed as IDE-integrated coding agents that decompose tasks…
6
RedCoder: Automated Multi-Turn Red Teaming for Code LLMs
自动化多轮红队测试框架，针对代码大模型的安全漏洞发起精准攻击。
arXiv:2507.22063v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) for code generation (i.e., Code LLMs) have demonstrated impress…
7
RAS: Measuring LLM Safety Through Refusal Alignment
arXiv新论文提出SafeVec白盒方法，革新LLM安全评估，从输出层转向拒绝对齐测量，更高效、更稳健。
arXiv:2606.25750v1 Announce Type: cross Abstract: Safety evaluation of large language models (LLMs) is commonly performed by querying models with unsa…
8
Giskard: LLM esting platform for preventing hallucinations and security issues
从多年实战经验出发，深入剖析2026年AI红队工具评估要点，助你防范LLM幻觉与安全漏洞。
Article URL: https://www.giskard.ai/knowledge/best-ai-agent-red-teaming-tools-in-2026-understanding-features-functions-and-solutions Comments URL: htt…
9
Safety Is Not Universal: The Selective Safety Trap in LLM Alignment
揭秘LLM安全评估的致命缺陷：看似全面防护实则暗藏针对特定群体的盲区，戳破“选择性安全陷阱”系统性风险
arXiv:2601.04389v3 Announce Type: replace-cross Abstract: Current safety evaluations of large language models (LLMs) create a dangerous illusion of un…
10
Evaluating LLMs for Real-World Web Vulnerability Detection
系统评估LLM在真实Web漏洞检测中的表现，揭示能力与局限，安全从业者必读
arXiv:2606.21397v1 Announce Type: cross Abstract: Large Language Models (LLMs) have emerged as a promising tool for automated vulnerability detection,…
11
Show HN: APK-interceptor – Android deeplink, Intent assessment helper
便携式Android测试APK，帮助安全工程师验证应用如何处理外部入口点如自定义URI方案和深度链接
Article URL: https://github.com/sterrasec/apk-interceptor Comments URL: https://news.ycombinator.com/item?id=48607766 Points: 1 # Comments: 0
12
Computational Safety for Generative AI: A Hypothesis Testing Perspective
用假设检验理论重新审视生成式AI安全问题，为模型风险量化提供严谨数学框架。
arXiv:2502.12445v2 Announce Type: replace Abstract: AI safety is a rapidly growing area of research that seeks to prevent the harm and misuse of front…
13
携程被罚 1000 万：未落实数据出境安全评估要求、违法出境个人信息
携程因未通过数据出境安全评估被罚千万，给所有出海企业敲响数据合规警钟
IT之家 6 月 13 日消息，据上海市互联网信息办公室，近期，在国家网信办指导下，上海市网信办针对属地部分企业网络数据安全主体责任履行不到位、安全管理防护措施缺失、后端处理数据合规能力不足、数据出境合规审计不严等问题办理了一批执法案件。 其中，针对上海携程商务有限公司未落实数据出境安全评估要求、违…
14
用“意外”衡量安全：Waymo 推出虚拟驾驶员，研究“老司机”如何应对紧急情况
Waymo 推出虚拟驾驶员模型 ReD，模拟人类如何应对紧急碰撞，为自动驾驶安全评估提供新视角。
IT之家 6 月 10 日消息，Waymo 今日在《自然 · 通讯》上发表了一篇研究论文，介绍了一个名为 ReD（Reference Driver，参考驾驶员）的计算机认知模型，旨在解释人类驾驶员如何在意外瞬间做出最有利的决策以规避碰撞。 该模型由 Waymo 与荷兰代尔夫特理工大学合作开发，公司希…
15
Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation
LLM从聊天走向自主行动，安全风险剧增：本文系统梳理智能体威胁面、攻击手法与防御评估体系。
arXiv:2606.10749v1 Announce Type: cross Abstract: Large language model (LLM) agents are rapidly moving from conversational interfaces to software comp…
16
When Behavioral Safety Evaluation Fails: A Representation-Level Perspective
从表示层揭示AI安全评估的盲点，为模型对齐提供全新视角与洞见
arXiv:2606.08044v1 Announce Type: new Abstract: Large Language Model (LLM) safety has often been evaluated at the behavior level, which provides limit…
17
VESTA: A Fully Automated Scenario Generation and Safety Evaluation Framework for LLM Agents
全自动生成测试场景，破解LLM Agent安全评估的痛点，附18页详实方法论与图表。
arXiv:2606.08531v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly evolving from simple text-based interaction systems into…
18
Safety officials finally have a good idea of what a big rocket explosion can do
火箭爆炸威力有了新标准，安全官员基于甲烷液氧火箭采用100% TNT当量评估，扩大安全区。
Overpressure from the Blue Origin blast shattered windows at a hangar about a mile away from the pad.
19
AICompanionBench: Benchmarking LLMs-as-Judges for AI Companion Safety
AI伴侣安全性如何评估？这个新基准让大模型自己当裁判，专治LLM越狱与不当行为。
arXiv:2606.04867v1 Announce Type: new Abstract: As AI companion platforms such as Replika and Character.AI rapidly grow, concerns about unsafe human-A…
20
A Multi-Domain Red Teaming Framework for Safety, Robustness, and Fairness Evaluation of Medical Large Language Models
首个针对医疗大语言模型的多领域红队框架，系统评估安全、鲁棒与公平性。
arXiv:2606.00027v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed across healthcare, yet existing benchmarks fa…