私人新闻 · 情报信号聚合
The November 12th-13th San Francisco program lists leaders from Intel, OpenAI, Thinking Machines Lab, Perplexity, Google, NVIDIA and AWS, as SGLang's serving framework continues to ship updates.
Epistemic status: reporting what OpenAI's own report and a Bitquery investigation say, plus my reading of it.
NVIDIA RTX Spark laptops are now available, bringing NVIDIA Blackwell RTX graphics, up to 128GB of unified memory and the full NVIDIA AI platform to slim, premium Windows PCs.
The composition of Anthropic’s workforce and the type of work getting done have changed significantly in the past year.
Can in-house chips pop the AI bubble? Microsoft's Maia 300 makes the case, but Nvidia shrugged off the news with a 1.29% decline.
Google is “starting to get onto that RSI treadmill”, cofounder Sergey Brin said in a roundtable on Gemini 4 Argon, Business Insider reported.
Anthropic president Daniela Amodei has warned that AI romantic partners risk driving people further apart, Business Insider reported.
Over the weekend, Masad clarified that OpenAI would replace Anthropic as its default model provider. Claude will still be available to customers on the platform.
Robots and AI models together could take on about 80% of the work Americans do, Anthropic says.
The US Federal Trade Commission has sent civil investigative demands to Anthropic and OpenAI, seeking documents about the safety incidents involving their AI models.
I’ll start with a nomenclatural sidenote: Apple seems to be going with “Apple HomeView” as the product’s name. E.g., the URL slug for it on Apple.com is apple-homeview/ , not just homeview/ . Few people notice and no one who doesn’t work in Apple product marketing cares, but they actually do the same thing with Vision Pro — in Cupertino, it’s “Apple Vision Pro” (once again including the URL slug ). Yet this week’s new HomePod Mini is just “ HomePod mini 2 ” and the existing 2nd-generation HomePo
OpenAI’s 110 staff in London received an average of about £2.3m each in 2025, counting salaries and shares.
协作游戏逆向工程网站 decomp.dev 宣布将关闭，管理员称 AI 让游戏反编译变得快且易于操作，基于人工的社区协作模式实际上已经瓦解，无法再区分哪些项目是以人工为主导的，哪些是由 AI 完成的。decomp.dev 管理员 Encounter 在官网声明中表示，从历史上看，匹配反编译一直是极其困难且耗时的任务，即便是对那些有适量活跃贡献者的中等规模游戏来说，反编译要以“年”为单位计时，但 GPT-6-Astra 之类的前沿 AI 模型最近已让反编译游戏变得极其轻松：它们在遵循复杂指令、为符号或调试信息交叉检索资源，以及在获得充分指引的情况下编写自然代码方面，要强得多。显而易见，只要有合适的指导、人类输入和代码审查，AI 能以非常显著的方式加速反编译。同样显而易见的是，反编译的社区属性已经被结束了。 https://decomp.dev/ https://games.slashdot.org/story/26/10/16/2057225/video-game-reverse-engineering-community-shuts-down-because-ai-has-made-it-too-easy
Attorneys general in Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia filed separate lawsuits against OpenAI on 14 October.
The public beta coordinates work through users' own accounts, while cloud model calls and Mac permissions put data access at the center of the product.
OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing.
天文学家可能发现了第一颗绿色恒星——名为 WD J1330+6435 的白矮星，位于地球 326 光年外。部分死亡恒星——即白矮星——在吞噬围绕其运行的行星后，可能会呈现出绿色。WD J1330+6435 一直在吸积其行星伴星上的岩石物质。这一过程使其大气中富含钙、锂和钠元素。模型表明这些元素的存在使得 WD J1330+6435 呈现绿色。WD J1330+6435 是已知首颗真正呈现绿色的恒星。研究人员称，白矮星大气中残留的钠元素吸收了光谱中的橙色部分，留下了幽灵般的绿光。 https://academic.oup.com/rasti/article/doi/10.1093/rasti/rzag069 https://www.space.com/astronomy/stars/astronomers-discover-the-1st-ever-green-star-a-cosmic-zombie-just-in-time-for-halloween
GNOME 项目已推出发行版项目 GNOME OS 一段时间，但它只提供基于最新代码的 Nightly 版本，主要面向开发者。现在它推出了面向普通用户的 GNOME OS Preview。GNOME OS Preview 基于 Nightly 版本，但去掉了 Nightly 特性，基于最新的 GNOME 51.1 构建。开发者表示该版本尚未经过充分的实际使用测试，可能会遇到问题，但稳定性肯定高于 Nightly 版本，对想要尝鲜的普通用户而言足够稳定，可用于日常使用。但它还需要进一步完善。 https://blogs.gnome.org/adrianvovk/2026/10/16/gnome-os-preview/
Redditor showcases a new quad 8-pin power-conversion kit for the 16-pin power connector on Nvidia GeForce RTX graphics cards.
Is the plug starting to be pulled on the GeForce GPU range? And is Nvidia turning to cloud gaming instead?
There is a newly leaked bill that Anthropic is purportedly shopping around is (if you assume its bugs get fixed).
上市即热销、用户排队催交付——这两条看似矛盾的主线，正同时发生在奕境品牌首款旗舰车型奕境X9身上。预售24小时小订破2万、上市24小时大定突破9157台，订单的高增长让“何时提车”成为用户社群最高频的提问；而品牌总经理曾清林直面用户的微博，则将这场交付攻坚战里的品牌取舍，真实呈现在公众面前。 “宁可让大家多等几天，多赔付超时积分，也绝不把一台没准备好的车交到用户手上。”这句承诺背后，恰恰藏着读懂奕境、读懂“境模式”的钥匙。 交付提速的表象之下，是品质标准的寸步不让 曾清林在微博中为交付定下了关键词：加速。值得注意的是，这份“加速”始终有一条不可逾越的红线——品质。品牌明确表示，交付提速不以牺牲品质为代价，既定验证流程、质量管控阀门一项都不会少。在行业普遍以交付速度为竞赛焦点的当下，奕境选择“速度为品质让路”，这种克制恰恰折射出其对产品与用户的长期主义。 作为「全栈原生共创」体系落地的首个战略成果，奕境X9不仅完整落地全栈华为乾崑智能科技，更导入了整套华为管控体系，以严苛验证守护产品的可靠性能与耐久品质。质量管控上，华为深度赋能23项整车验证标准与方法，多项指标跻身行业第一梯队。双方联合搭建自动化测试实验室，通过“测试前移”与自动化压力测试提前识别和控制风险，覆盖整车4000多个接口，累计压测超40万次；针对方坑、斜坡坑、飞坡等30多种典型路况，设计220多种不同场景，远超行业标准。 “境模式”的本质：不是装上去，而是长出来 理解奕境的交付底气，需要回到合作的源头。奕境是东风与华为乾崑全栈原生共创的全新品牌，双方围绕研、产、供、销、服全价值链深度共创，以1:1深度绑定、长期共生、持续投入，构建「全栈原生共创」体系，并将共创合作拓展至用户端与供应链端，实现全产业链价值共生的完整生态，业界称之为”境模式”。 它与市场上常见的”单点合作”有着本质区别。常规合作模式下，华为往往以供应商身份提供某一部件或系统，车辆在既有架构上装配智能能力；而在“境模式”下，华为从车型定义阶段便全程参与，将乾崑智驾ADS、鸿蒙座舱等全栈技术深度嵌入整车原生架构，双方以同一套技术语言共同定义产品。 这种深度融合带来的最直接成果，是产品力的系统性领先。华为乾崑最新、最全面的智能技术，已在奕境品牌实现全系搭载：首款车型奕境X9落地最新一代华为乾崑全栈方案，涵盖华为乾崑智驾ADS 5、鸿蒙座舱6、华为乾崑赤兔平台、车载光、车云及鲸鳍通信。这些技术不是外挂的功能点，而是融入整车基因的原生能力，双方共同打造的天穹智盾安全架构，更代表行业最高安全标准，这正是”境模式”区别于行业其他合作形态的核心成果。 东风×华为：国家队底蕴与科技实力的双向奔赴 “境模式”之所以能成为行业最优合作范式，根源在于合作双方的不可替代性。东风作为央企国家队，积淀成熟的智能制造体系、严苛的品控管理能力与数十年整车制造经验；华为则手握智能汽车时代领先的全栈技术能力，建立起一套高规格的研发验证标准。 双方强强联合，赋予“境模式”两大难以被复制的核心能力。一是“造车的高度”，依托东风与华为的协同创新，双方将联合研发成果转化为产品核心竞争力。在央企制造功底与华为双重质量标准共同加持之下，奕境成都工厂对每一台整车执行多轮下线检测，守住整车交付品质底线，最终为用户带来高品质智慧出行体验；二是“智能的深度”，依托全栈原生共创机制，华为乾崑全套智能技术深度融入整车开发，技术服务于真实用车体验，而非单纯堆砌参数指标。央企深厚造车积淀叠加科技企业顶尖技术实力，构筑起“境模式”难以逾越的竞争壁垒。衡量一种合作模式的成色，最终要落到产品与用户身上。首批提车的奕境X9车主已在社交平台晒出实拍，一句”值了”，是对这款车、也是对背后这套合作模式最真实的褒奖。 当行业还在争论该以何种姿态拥抱华为时，奕境给出了自己的答案：真正好的合作模式，不是技术移植的深浅，而是能否让技术长出产品的灵魂，让品质守住用户的信任。从这个角度看，奕境汽车董事长汪俊君“最好的华为在奕境”的表态，并非一句口号，而是一种经得起市场与时间检验的选择。奕境模式，在为千万中国家庭带来高品质出行体验的同时，也为国内智能汽车产业转型升级提供了一份可参考的实践样本。
Anthropic's AI model Claude has solved a 25-year-old cryptography puzzle. The mysterious seeds behind NIST's standard curves for digital signatures were provided by the NSA without explanation, raising long-standing suspicion about built-in backdoors.
Glenn Fleishman: I wondered what the entry-level, initial price of each Apple TV box at introduction was, from the March 2007 original to the Apple TV 4K (4th generation) announced October 13, 2026, adjusted for inflation. So I charted it. I suppose this doesn’t mean Apple has given up on making Apple TV content accessible, but the price-accessible path is just using the Apple TV on the crapware OS your “smart” TV runs. Link: glog.glennf.com/blog/2026/10/17/apple-tv-price-history/
Anthropic's robotics research puts more hands-on jobs in AI's path. But messy tasks such as gardening remain hard to automate.
“Pre-LLMs a Chrome full chain would take a team of four around 6-9 months to be produced, our first submission took 2 months.
“Our AI Office has started enforcing these rules, and sent Requests for Information addressing issues such as security, external evaluation and monitoring how models are used,” EU tech chief Henna Virkkunen said in a speech in Dublin on 14 October.
近日，微信鸿蒙版已上线小艺语音操作能力。唤醒小艺后，直接说“用微信给XX发消息”“给XX打微信语音”或“给XX打微信视频”，即可完成对应操作，发送消息还可直接说出具体内容。如遇联系人重名或指令不清，小艺会提示确认。开车、做饭、做家务等不便使用双手时，沟通更加简单直接。更新微信鸿蒙版和小艺至最新版本即可体验。
Autonomous agents are the hottest thing in AI. OpenAI, Meta, Google, Instinct, and others promise their agents can scour the internet on your behalf, doing anything from booking a dinner reservation and cleaning out your inbox to buying new shoes.
“Trail of Bits has committed their entire security research organization,” OpenAI wrote in June, when the two companies launched a joint project to fix open-source software.
左右滑动切换「今天 / 全部 / 星标」