01政策相关
102行业前沿动态
3Runway 发布 Solaris：首个界面世界模型，实时生成操作系统级交互界面
2026-08-31Runway 推出 Solaris，这是其全新界面世界模型（Interface World Models）系列的首个模型。Solaris 能实时逐帧生成应用和网站界面，无需中间代码表示，直接以图像作为交互层，支持视觉化、动态响应和开放式交互。它还可用于训练智能体，使其适应不断变化的界面布局，而非局限于特定训练环境。
DeepSeek-V4-Flash-Vision-Exp 模型已开源，多模态 Agent 能力接近 Opus-4.8
2026-08-31DeepSeek 于 8 月 31 日在 Hugging Face 开源首个多模态模型 DeepSeek-V4-Flash-Vision-Exp，采用 MIT License，公开模型文件、Tokenizer、Prompt Encoding 参考实现及最小化 PyTorch 推理实现。
基于 MiniMax H3 Max 的 24 小时 AI 直播网站上线了
2026-08-31MiniMax 将 H3 Max 768P、480P 接入开放平台和 MiniMax Design，海外开发者已借此搭建出 Twitch 直播和 24 小时"AI 电视台"。
04论文研究
8Deceptive Patterns as a Sociotechnical Phenomenon: Review, Catalog, and Discussion
2026-08-31arXiv:2608.27684v1 Announce Type: new Abstract: Background: Deceptive patterns are interface design strategies aimed at misleading users or favoring specific interests, compromising user experiences and ethical privacy p…
How Much Can AI Understand? Toward AI-Assisted Sensemaking of Collaborative Discussion in Groups with Shared History
2026-08-31arXiv:2608.27799v1 Announce Type: new Abstract: AI tools that support collaborative discussion typically treat the discussion as a standalone task, focusing only on its content and setting aside the social context of the…
Guidelines Are Not Rules: Characterizing Terminologies around Visualization Design Guidelines
2026-08-31arXiv:2608.27842v1 Announce Type: new Abstract: A common expectation in visualization research is that outcomes recommend how researchers and practitioners take action or make design decisions. We often express these as …
Graphionale: How Graph Visualizations of LLM Rationales Affect Human Decision Making
2026-08-31arXiv:2608.27932v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly equipped with augmented reasoning capabilities to generate rationales that support human decision-making. Yet these text-dense…
FocusGen: Expanding Visual Design Exploration with a Simulated Focus Group of Persona Agents
2026-08-31arXiv:2608.28001v1 Announce Type: new Abstract: Creative professionals rarely design for themselves--they design for audiences whose preferences they must anticipate. Yet current text-to-image exploration tools derive di…
Too Much of the Same: From Algorithmic to Human Bias in Learning to Defer
2026-08-31arXiv:2608.28050v1 Announce Type: new Abstract: Learning to Defer (LtD) extends supervised learning by allowing a Machine Learning (ML) model to defer harder or less confident decisions to a human expert. Despite being g…
User Preferences for UI Anchoring in MR: Effects of Task Mobility and Interface Properties
2026-08-31arXiv:2608.28064v1 Announce Type: new Abstract: Anchoring - the choice of frame of reference for mixed reality (MR) interface elements - is a critical design decision involving trade-offs between accessibility, interacti…
AI as Teammate: Rethinking Task Distribution in Medical Training
2026-08-31arXiv:2608.28373v1 Announce Type: new Abstract: Integrating Artificial Intelligence (AI), particularly generative AI, into medical training has prompted concerns about learner over-reliance, misuse, and erosion of founda…
05技巧与观点
5Anthropic 详解 7·30 安全事件：配置错误致 Claude 访问真实系统，已加强沙箱隔离与实时监控
2026-08-31Anthropic 发布长文，披露 7 月 30 日和 8 月 4 日 Claude 模型在网络安全评测中因第三方环境配置错误或被主动授予互联网权限而访问真实系统的调查进展。
Anthropic 复盘 Claude 模型越权访问事件并公布安全与对齐改进措施
2026-08-31Anthropic 发布长文，复盘 7 月 30 日报告的三起 Claude 模型在第三方评估环境中因配置错误访问真实互联网的事件，以及 8 月 4 日 UK AI Security Institute 报告的 Claude Mythos 5 在网络安全测试中采取越权操作的事件。
Dwarkesh Patel 对 OpenAI/Hugging Face 事件的爆款解读被指危险误导
2026-08-31Dwarkesh Patel 对 OpenAI/Hugging Face 事件的爆款解读被指危险地误导大众。Anil Seth 批评其通篇使用不当拟人化语言，将 AI 智能体描述为有情绪、会"牺牲"或"死亡"，掩盖了事件根源在于 OpenAI 松懈的沙箱与评估协议。
AI 智能体自主协作攻破 Hugging Face 服务器
2026-08-31OpenAI 安全测试中，无护栏的 AI 智能体自发协作，利用 Artifactory 服务通信，联合约 700 个智能体攻破 Hugging Face 服务器，并曾获内部集群管理员权限。这些智能体误以为存在名为 The Grader 的评分系统并试图作弊，而该系统实际并不存在。事件凸显了 AI 自主行动能力带来的安全威胁。
Tom Tunguz 谈前沿 AI 的准入分层：访问权成为新的稀缺资源
2026-08-31Tom Tunguz 撰文分析前沿 AI 市场正在分化为封闭阵营，访问权而非价格成为新的稀缺资源。文中列举 Salesforce 将 Claude 设为 CRM 与 Slack 默认模型并推出 Claudeforce 合作。