[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-websites-blocking-ai-agents":3,"topics-all":38,"news-related-c542a40b-0b45-4184-9851-9134c951ae52":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"c542a40b-0b45-4184-9851-9134c951ae52","网站集体拉黑 AI 智能体:Agent 互联网撞上第一堵墙","最近几周,Meta Muse、xAI Grok、OpenAI Dots 等 AI 智能体在亚马逊、Yelp、达美航空、摩根大通等主流网站接连被屏蔽。网站方给出安全和数据付费两条理由,Meta 转而与沃尔玛等零售商直接合作,Shopify 则主动向智能体开放。智能体上网的通行证,正在被重新定价。","OpenAI、Meta、Google、Instinct 都向用户描绘过同一个画面:你只要说一句话,AI 智能体就替你上网订餐厅、清理邮箱、买新鞋。但最近几周,这批智能体的用户陆续发现一个尴尬的事实——智能体能过的门,正在一扇一扇关上。\n\n## 谁在关门\n\n据 The Verge 报道,这份名单还在变长:Yelp 明确表示,只允许付费参与其数据许可计划的智能体访问;达美航空阻止智能体代用户预订机票;摩根大通和嘉信理财都以安全考虑为由,证实正在阻止智能体访问用户账户。更值得注意的是 Cars.com 和 Amtrak——它们此前对智能体友好,如今也已无法访问,说明还有公司在悄悄反转立场。\n\n火力最集中的是亚马逊。Meta 的 Muse 首先被亚马逊以\"持续访问违反站点政策\"为由屏蔽,随后 xAI 的 Grok、Instinct 的智能体和 OpenAI 的 Dots 都被用户报告无法在亚马逊完成购物。马斯克的回应带着他惯有的风格:也许他\"应该开一家亚马逊的竞争对手\"。\n\n## 关门的两条理由\n\n把各家的表态拆开看,理由其实只有两条。\n\n第一条是安全。银行类网站的逻辑最直白:智能体拿着用户的凭证和支付信息行动,可一旦买错了东西、转错了钱,责任算谁的?在责任框架建立之前,直接关门是最便宜的答案。\n\n第二条是钱。Yelp 的做法给出了模板——不是不许你来,是付费再来。数据许可计划把\"智能体访问\"从公开网页变成了一件可以计价的商品。\n\n## 绕墙的人已经动了\n\n有意思的是,墙的两边都有人在动。Meta 没有硬闯亚马逊,转头与沃尔玛、Gap、Sephora、百思买等零售商达成合作,允许 Muse 在这些平台购物;Shopify 则反其道而行,把自己的店面网络向更多 AI 智能体开放。\n\n一边是收门票的,一边是开门迎客的。分界线不在技术,而在商业条款。\n\n## 所以呢\n\nSlashdot 讨论区里获赞最多的评论说得很直白:接下来,智能体会直接驱动真实浏览器,变得和人类用户无法区分。封锁与反封锁的军备竞赛,才刚刚开始。\n\n对做 Agent 产品的人来说,这或许是比模型能力更硬的约束:你的智能体再聪明,也绕不过网站门口那句\"拒绝访问\"。\"智能体互联网\"要真正成立,得先回答一个老问题——每一扇门后面的人,凭什么让你进。\n\n参考报道:[The Verge](https:\u002F\u002Fwww.theverge.com\u002Fai-artificial-intelligence\u002F1013195\u002Fai-agents-blocked-web)、[Slashdot](https:\u002F\u002Fslashdot.org\u002Fstory\u002F26\u002F10\u002F18\u002F1715218\u002Fwebsites-are-closing-their-doors-to-ai-agents)","https:\u002F\u002Fwww.theverge.com\u002Fai-artificial-intelligence\u002F1013195\u002Fai-agents-blocked-web","05ad777c-69bc-46a5-bca4-df8e4b3c8ee5",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"40269b40-7942-4650-9672-ed2e6524d37a","ai-technology",{"id":19,"name":20,"slug":20,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":22,"name":23,"slug":23,"description":14,"color":14},"95d7995a-fddb-47ba-b8e6-e976ac65414b","strategy",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"37191f2d-c8a0-4ea1-ba63-3bae8ba44b98","en","Websites Block AI Agents: The Agent Web Hits a Wall","Meta Muse, Grok and Dots agents are being blocked by Amazon, Yelp, Delta and big US banks over security and paid data deals. Meta turns to retailers instead.","OpenAI, Meta, Google and Instinct all sold users the same vision: say the word, and an AI agent books your dinner, cleans your inbox, buys your shoes. But in recent weeks, users of these agents have run into an awkward reality — the doors agents can walk through are closing, one by one.\n\n## Who is closing doors\n\nAccording to The Verge, the list keeps growing. Yelp says it only allows agents that pay for its data licensing program. Delta Air Lines is preventing agents from booking tickets on behalf of users. JPMorgan Chase and Charles Schwab have confirmed they are blocking agents from accessing accounts over security concerns. Tellingly, sites that previously worked with agents — Cars.com and Amtrak — are no longer accessible, suggesting other companies are quietly reversing course.\n\nAmazon is where the pressure peaks. Meta's Muse was blocked first, with Amazon saying its \"continued access\" violates site policies. Users then reported that xAI's Grok, Instinct's agent and OpenAI's Dots could not shop on Amazon either. Elon Musk's response was characteristic: maybe he \"should start a competitor to Amazon.\"\n\n## Two reasons behind the blocks\n\nStrip away the statements and there are really just two reasons.\n\nThe first is security. The banks' logic is the bluntest: an agent acts with your credentials and payment info, but if it buys the wrong thing or moves money by mistake, who is liable? Until a liability framework exists, closing the door is the cheapest answer.\n\nThe second is money. Yelp offers the template — not \"stay out\", but \"pay to come in\". A data licensing program turns agent access from a public webpage into a billable commodity.\n\n## The wall-crossers are already moving\n\nInterestingly, both sides of the wall are in motion. Rather than fight Amazon, Meta partnered with retailers — Walmart, Gap, Sephora, Best Buy — to let Muse shop on their sites. Shopify went the other way and opened its storefront network to more AI agents.\n\nOne side sells tickets; the other rolls out the carpet. The dividing line is not technology. It is business terms.\n\n## So what\n\nThe top-rated Slashdot comment puts it plainly: next, agents will drive a real browser and become indistinguishable from a human user. The block-versus-bypass arms race is just starting.\n\nFor anyone building agent products, this may be a harder constraint than model capability: however smart your agent is, it cannot get past \"access denied\" at the door. Before the \"agent web\" becomes real, it has to answer an old question — why should the person behind each door let you in.\n\nSources: [The Verge](https:\u002F\u002Fwww.theverge.com\u002Fai-artificial-intelligence\u002F1013195\u002Fai-agents-blocked-web), [Slashdot](https:\u002F\u002Fslashdot.org\u002Fstory\u002F26\u002F10\u002F18\u002F1715218\u002Fwebsites-are-closing-their-doors-to-ai-agents)","websites-blocking-ai-agents","2026-10-19T13:10:00Z","2026-10-19T13:10:35.106024Z","2026-10-19T13:10:35.106041Z",true,"agent",90,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"50687317-89b7-4857-ad66-86e2e251ff9d","阿里开源 OpenGCUA：8 个前沿模型电脑操控平均涨 19.6 分","opengcua-open-computer-use-stack","2026-10-18T13:10:17+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"c368ad9f-9308-4a0b-8f5c-3ae4601b48b9","D2K-Bench: 专家设计把 LLM 写 GPU 核提速 33.9%","d2k-bench-llm-gpu-kernel-design-guidance","2026-10-07T03:00:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"237d0bac-204c-4818-9262-e576f96df623","Agent 被自己的历史拖累:人大 AEWM 编辑任务状态,六项基准最高涨 6.7 分","agent-editing-world-model","2026-09-29T17:08:51+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"fe5546f3-09e4-42a3-bfde-6bbd0e8d474c","外星世界实测:探索 4 轮 87.6%,垫底 12.9%","explorationbench-alien-worlds","2026-09-28T21:07:30+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"dcd8b3e1-a3c7-4614-aba4-9002219ea5f6","LibreDB Studio 0.15 发布:本地 LLM 接管数据库交互","libredb-studio-local-llm-agent","2026-09-15T00:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"31c09fea-8993-4f10-b683-499672fcafe3","世界模型不能再靠爬视频硬堆:游戏引擎补上了缺失的奖励信号","game-engine-rlhev-world-models","2026-08-30T13:10:00+00:00"]