今日必看

Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery

arXiv:2609.09647v1 Announce Type: new Abstract: Agentic systems are rapidly moving to production, where they read untrusted inputs, call tools with real permissions, and act autonomously, expanding the security surface beyond chat-only models. Yet standard evaluations remain sing

CGSM: Concept-Guided Segmentation Model for Precise Pulmonary Lesion Delineation

arXiv:2609.07004v2 Announce Type: replace Abstract: Accurate segmentation of pulmonary lesions is essential for effective clinical diagnosis and treatment strategies. Existing segmentation approaches often lack task-specific semantic guidance, as text-based annotations typically

CHIMERA Challenge Task 2 and 3: Response Subtypes Classification and Progression Survival Prediction in Bladder Cancer Patients using Multimodal Datasets

arXiv:2609.09510v1 Announce Type: cross Abstract: High-risk non-muscle-invasive bladder cancer (HR-NMIBC) carries substantial risks of recurrence and progression, while current clinical risk stratification remains limited. CHIMERA was established as a multimodal AI challenge to b

Can AI Agents Deliver Verifiable Network-Wide Outcomes Across Authority Boundaries?

arXiv:2609.10181v1 Announce Type: cross Abstract: AI agents are increasingly involved in network automation, where they can initiate configuration changes through mediated operational interfaces and assess the resulting state. Nonetheless, operational networks usually span many d

DeepSeek 发布 V4.1-Flash，大幅降低 AI Agent 的 KV cache 内存需求

DeepSeek 发布多模态模型 V4.1-Flash，以 MIT 许可开源在 Hugging Face，目标是大幅压缩 KV cache 和长上下文处理成本。 🔗 阅读原文 via AIHOT · https://aihot.news/items/cmtvj9kae0aq1ronbfyxuewfb

Rebuilding AUTOMATIC1111 with Gradio Workflow

Sep 10, 2026Frontier Red TeamMeasuring tactical intelligence targeting and conventional weapons capabilities of AI models

Sep 10, 2026Frontier Red TeamMeasuring tactical intelligence targeting and conventional weapons capabilities of AI models

T. Rowe Price brings more of Claude to its investment process

T. Rowe Price brings more of Claude to its investment process

What 1,000 small business owners taught us about AI

What 1,000 small business owners taught us about AI

Muse can shop, write emails, and negotiate prices for users, all through WhatsApp

Meta unveils Muse, an AI agent that books travel, handles purchases, and sends emails through WhatsApp, complete with a payment feature that runs through Stripe's Link. That puts Meta ahead of OpenAI, which stopped its direct checkout feature in ChatGPT. A separate security agent

npm extends recovery-code security holds to all accounts

npm now places a temporary 72-hour security hold on any account after a successful recovery-code sign-in, extending a protection that previously applied only to high-impact accounts. This change applies to all npm accounts. During the hold, publishing and other security-sensitive

CodeQL 2.27.0 adds support for Linux ARM64

CodeQL 2.27.0 is now available on Linux ARM64, adds a new Rust security query, expanded framework coverage for Java/Kotlin and C#, and analysis accuracy improvements across multiple languages. CodeQL is the static analysis engine behind GitHub code scanning, which helps you find

GPT-6 Astra: The next generation in intelligence for work

Meet GPT-6 Astra, OpenAI’s most capable model for business, with advanced reasoning, computer use, and stronger writing and design judgment.

I made an MCP app so claude code can record, edit and export demo screen recordings for you

Just say "film a 30-second demo of my app" and your agent opens it, moves the cursor, clicks through it, and hands you back a polished video. Works with Claude Code, Codex, OpenClaw or any compatible AI agent through MCP. Supports basically every app or website! Just ask and get

Frigade Assist API

Give your AI agent guides that show users where to click Discussion | Link

#716.Stripe：企业 AI 的关键是治理

📌 一句话摘要 Stripe 工程经理 Sharadh 拆解公司级 Agent Kai 的构建逻辑：企业 AI 的难点不是提供工具而是治理，靠 Project 划定权限与成本边界、靠 skill 平台把一次性对话变成可复用工作流，不到 10 人团队服务超 1 万名员工。 📝 详细摘要 本期 How I AI 由主持人 Claire Vo 对话 Stripe 工程经理 Sharadh，拆解内部公司级 Agent Kai 的构建与治理实践。Kai 不是聊天工具，而是一套结合个性化上下文、组织架构、数据平台、沙盒、skill 与治理机制的企业级 AI 系统，每

AI agents are flooding public services with new requests

“The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing,” the researcher told TechCrunch.

破除人形崇拜！京东用“狼族”军团重做物理AI

Castle Labs：Fomo 近一周日均交易量约 3.5 亿美元

📌 一句话摘要 Castle Labs 披露社交交易应用 Fomo 已成为加密市场最大交易终端，占据该赛道约 40% 份额，近一周日均交易量约 3.5 亿美元。 📝 详细摘要 吴说快讯援引 Castle Labs 披露的数据称，社交交易应用 Fomo 目前已成为加密市场规模最大的交易终端，占据该赛道约 40% 的市场份额；过去一周内其日均交易量（Average Daily Volume）约为 3.5 亿美元。内容为单一数据点的转述，未提供统计口径、数据来源方法或与其他终端的对比。 💡 主要观点 Fomo 成为加密市场最大交易终端，占该赛道约 40% 份额

The Frontier AEO Tracker: What Astra Chooses (and every other frontier model, and what you can do about it)

Naive autoresearch investment in our AEO have yielded impressive ROI, and so naturally it was time to take it seriously. We were inspired by What Claude Code Actually Chooses , and decided to extend/adjust it to our tastes. After a few billion tokens of prototyping, aligning, and

Video compressor

Tool: Video compressor I recorded a short demo video of my Equal Earth animation on my phone and wanted to publish an optimized version of that video (using FFMPEG) on my blog, so I had Claude Fable 5.1 in Claude Code for web build me this tool using the WebAssembly build of FFMP

A catwalk at IFA, but for robots

PLUS: Figure's humanoid play gets a compute boost

Show HN: SiteTweak – a browser extension to modify any website

I've made an extension that works in the sidebar and lets you modify any website with AI. I tried to make it feel like using Cursor or Codex. You can talk to it in a chat panel to refine the modification, ask a question, or request automation. If you'd like, you can plug in your