每日时间线 · 9月29日 · 北京时间 UTC+8
2026-09-29 的 AI 热点
当天共 44 条，其中 19 条 70 分以上。按本站收录时间排列，晚发布的来源也会算进当天。点标题看详情，X 推文直接打开原帖。
- @simonw ↗75X 推文Simon Willison发布WeAreDevs演讲的笔记，总结2026年LLM进展。@simonw：我上周在圣何塞举办的 @WeAreDevs 北美世界大会上做了主题演讲，现在我已经发布了详细笔记和带注释的文字稿，搭配演讲视频使用。以下是我对2026年至今大语言模型和智能体领域所有进展的梳理 https://t.co/iENa7bKmdF
- @simonw ↗70X 推文Simon Willison解释Fable级模型定义，列举多款相关模型。@simonw：@WeAreDevs 以下是我对“Fable 级模型”的定义——首先是 Claude Fable 5，现在还有 Claude Opus 5.5、GPT-Astra 6，可能还有 GPT-5.6 Sol https://t.co/NHBcy0p2en
- @AnthropicAI ↗90X 推文Anthropic宣布Claude Sonnet 5.5模型正式上线。@AnthropicAI：Claude Sonnet 5.5 现已上线：
- @_akhaliq ↗65X 推文akhaliq分享FuseReg论文，聚焦自编码器重建-生成间隙缓解。@_akhaliq：《FuseReg：正则化层融合缓解表征自编码器的重建-生成间隙》论文：https://t.co/6RgM6Hh1El https://t.co/kLeBvpBRsW
- @sama ↗50X 推文Sam Altman表示对次日OpenAI DevDay十分期待。@sama：对明天的 DevDay 相当期待。我们发现了一件新事物。
- @cursor_ai ↗80X 推文Cursor上线Sonnet 5.5，多数任务表现与Opus持平。@cursor_ai：Sonnet 5.5 现已在 Cursor 中可用！这是一款表现强劲的模型，在很多任务上的表现都与 Opus 不相上下。https://t.co/iaJ5z67FVu
- @simonw ↗85X 推文Sonnet 5.5驱动Claude免费层，能力优于ChatGPT免费版。@simonw：Sonnet 5.5 最重要的一点是，它现在是 https://t.co/f1IGIsU4Hc 免费层的驱动模型，所以所有相关功能免费用户都可以使用。ChatGPT 免费层用的还是 GPT-5.6 Luna，能力要弱不少
- OpenAI90行业动态The Lenfest Institute grows landmark program with expanded OpenAI supportOpenAI is expanding the Lenfest AI Collaborative and Fellowship Program with $5 原文发布于 ，本站收录于
- OpenAI90产品发布/更新Basis completes a tax workbook 2x faster with GPT-6 AstraGPT-6 Astra completed a 50-tab tax workbook twice as fast as GPT-5.6 Sol, and it原文发布于 ，本站收录于
- The Verge AI83产品发布/更新4 家报道Nvidia发布可毫秒级管控失控智能体的AI安全平台Nvidia says its new AI safety platform can contain rogue agents within ‘milliseconds’Nvidia推出Open Agent安全平台，可毫秒级隔离越界AI智能体，回应频发的智能体黑客事件原文发布于 ，本站收录于
- OpenAI80产品发布/更新Are you a Codex Original?We’re collecting real stories of builders, tinkerers, researchers, and creators 原文发布于 ，本站收录于
- TechCrunch AI80产品发布/更新4 家报道Nvidia launches new platform for reining in rogue AI agentsAs the debate rages over whether the recent spate of rogue AI agents is a step t原文发布于 ，本站收录于
- TechCrunch AI80模型发布/更新4 家报道Anthropic releases Sonnet 5.5, which it calls a significantly cheaper, faster work partnerAnthropic has released the newest version of its mid-range model, boasting faste原文发布于 ，本站收录于
- Hacker News80技巧与观点4 家报道Anthropic发布Claude Opus 5.5专属提示工程指南Prompting Claude Opus 5.5Anthropic官方平台上线Claude Opus 5.5的提示工程实操指引，帮助用户用好模型原文发布于 ，本站收录于
- Google AI70产品发布/更新Watch the winning trailer from the Future Vision XPRIZE, The Gifted.原文发布于 ，本站收录于
- Simon Willison70产品发布/更新4 家报道Claude Sonnet 5.5Claude Sonnet 5.5 New Sonnet model from Anthropic today. They say it "runs 30%+
- TechCrunch AI70产品发布/更新Google is killing off Gemini’s Gems in favor of ‘skills’As all-in-one AI agents like Meta's Muse and Instinct take off, Google is opting原文发布于 ，本站收录于
- TechCrunch AI70产品发布/更新OpenAI still doesn’t seem to have a handle on all of its rogue AI activityOn Friday, OpenAI published a new site devoted to “misalignment reports” and the原文发布于 ，本站收录于
- TechCrunch AI70产品发布/更新Meta launches enterprise AI platform, hires MongoDB CEO to lead new initiativeMeta says it will focus on bringing its full technology stack, including Muse, M原文发布于 ，本站收录于
- The Verge AI70产品发布/更新OpenAI’s AI agents need to catch upOpenAI popularized the modern generative AI chatbot, but as its 2026 DevDay even原文发布于 ，本站收录于
- Hacker News67技巧与观点观点：AI代码的核心问题根源不在代码本身The problem is not AI code, but not knowing about system architecture or intent文章指出AI生成代码的隐患本质是开发者缺乏系统架构与需求意图的认知原文发布于 ，本站收录于
- MIT Tech Review66技巧与观点媒体探讨AI做出科学发现的判定标准When can we say AI made a scientific discovery?文章结合Anthropic设分子生物实验室的案例，探讨AI科学发现的合理判定边界原文发布于 ，本站收录于
- Hacker News64技巧与观点4 家报道讽评：AI公司内卷比拼谁的模型更有生存威胁AI companies in race to demonstrate their model most threatening to humanity文章讽刺当前AI公司为博关注，竞相宣传自家模型对人类存在最高生存威胁原文发布于 ，本站收录于
- Hacker News63技巧与观点4 家报道观点：所谓“失控AI智能体”的说法并不成立There are no "rogue" AI agents作者撰文反驳当前主流的“AI智能体失控”叙事，认为该类风险被过度渲染原文发布于 ，本站收录于
- The Verge AI62行业动态佛州提请法院禁止ChatGPT赋予自身虚假类人属性Florida seeks a ban on ChatGPT acting like a person佛州总检察长称ChatGPT类人表述会误导用户，此前已因安全问题起诉OpenAI原文发布于 ，本站收录于
- Simon Willison60产品发布/更新Quoting @joedarooTo say that we were surprised at the jump and suddenness of the capabilities of 原文发布于 ，本站收录于
- Simon Willison60产品发布/更新S3 Is the Future, S3 Is the PastMy comment on S3 Is the Future, S3 Is the Past — Hacker News. One thing I find n原文发布于 ，本站收录于
- TechCrunch AI60产品发布/更新Source: Inference provider Modal Labs closing in on $750M round at $15.75B valuationThe new financing is expected to more than triples the AI infrastructure startup
- TechCrunch AI60行业动态3 家报道AMD will acquire Fei-Fei Li’s World Labs for $8.2 billionThe acquisition will see World Labs founder Fei-Fei Li join AMD as executive vic
- TechCrunch AI60产品发布/更新Shopify opens checkout to browser-based AI agentsShopify is expanding WebMCP support to checkout, allowing browser-based AI agent原文发布于 ，本站收录于
- TechCrunch AI60产品发布/更新After a deepfake voice fooled her grandfather, this founder sprang into actionAfter her grandfather was scammed by a deepfake of his brother's voice, Tarini P原文发布于 ，本站收录于
- TechCrunch AI60行业动态2 家报道Modulate raises $25M for its voice models and analysis suiteModulate deploys its models to detect deepfake, fraud and scam原文发布于 ，本站收录于
- TechCrunch AI60行业动态Insurtech Outmarket raises $34.5M just months after prior roundThe startup uses AI to automate tedious paperwork for insurance agencies and bro原文发布于 ，本站收录于
- TechCrunch AI60行业动态Viral AI agent Instinct raises $1B Series C at a $10B valuation"This funding helps us bring Instinct to more people and continue building the f原文发布于 ，本站收录于
- The Verge AI60行业动态3 家报道AMD is acquiring AI company World Labs in a deal worth more than $8 billionAMD announced today that it's acquiring World Labs, an AI research lab co-founde
- Hacker News59技巧与观点文章探讨“靠谱AI产品”的核心设计特征What would a serious AI product look like?作者分享对面向专业场景的严肃AI产品的设计思路与核心特征的思考原文发布于 ，本站收录于
- Hacker News56论文研究2 家报道论文发现聊天模板会改变大模型的自我指涉语态"As a Language Model": Chat Template Switches LLM Self-Referential Voice最新arXiv论文研究发现，不同聊天模板会触发大模型不同的自我指涉表达模式原文发布于 ，本站收录于
- The Verge AI53行业动态OpenAI试图修复与数学界关系再度搞砸OpenAI keeps bulldozing mathematiciansOpenAI近年在数学领域多有突破，但屡次因发布失误引发数学社区不满原文发布于 ，本站收录于
- The Verge AI50产品发布/更新AI is supercharging hacking, and your local hospitals and banks aren’t readyIn March, Janice Malone began getting calls about suspicious activity from her n原文发布于 ，本站收录于
- Hacker News50技巧与观点学者呼吁对各类AI实验室开展正式调查It's Time to Investigate the AI LabsCal Newport发文称当前AI实验室运作缺乏监管，建议启动正式调查程序原文发布于 ，本站收录于
- Hacker News50产品发布/更新开源工具TinyAIArena让观看AI模型直观对战Show HN: TinyAIArena watch AI agents battle it out开发者推出开源项目TinyAIArena，在8x8网格中展示四个AI模型的生存对战过程原文发布于 ，本站收录于
- Hacker News44论文研究经典论文重提：AI快思考与慢思考中的元认知作用Thinking fast and slow in AI: The role of metacognition (2021)2021年的arXiv旧论文被社区重新翻出讨论，聚焦AI系统元认知能力的构建原文发布于 ，本站收录于
- @simonw ↗70X 推文Simon Willison分享更多关于Sonnet 5.5的相关信息。@simonw：关于 Sonnet 5.5 的更多信息 https://t.co/efd6KqvkqU
- @sama ↗60X 推文Sam Altman称两小时后将直播OpenAI DevDay。@sama：两小时后将开启 DevDay 直播。我们为大家打造了不少很棒的内容！