Meshy 跻身 a16z 消费级 AI 应用月收入 Top 50，为榜单唯一 AI 3D 公司
在 a16z 首份消费级 AI 应用月收入榜单中，Meshy 位列第 31 名，与 OpenAI、Anthropic、Canva、Superhuman、Higgsfield 等共同上榜
#人工智能
外媒与科技媒体长文、研究号文章
量子位利好AI 评分3535 彭博AI 评分1616 DeepSeek Nears $12B Funding Round Ahead of Potential IPO
“Bloomberg: The China Show” is your definitive source for news and analysis on the world's second-biggest economy. From politics and policy to tech and trends, Yvonne Man and Minmin Low give global investors unique insight, delivering in-depth discussions with the newsmakers who matter. (Source: Blo
雷锋网AI 评分3333 Meshy 跻身 a16z 消费级 AI 应用月收入 Top 50，为榜单唯一 AI 3D 公司
在 a16z 首份消费级 AI 应用月收入榜单中，Meshy 位列第 31 名，与 OpenAI、Anthropic、Canva、Superhuman、Higgsfield 等共同上榜加州硅谷，2026 年 10 月 5 日 —— 全球领先的 AI 3D 多模态模型公司 Meshy 今日宣布，公司入选由硅谷风险投资机构 Andreessen Horowitz（a16z）发布的“消费级 AI 应用月收入Top 50”榜单。Meshy 位列第 31 名，是榜单上唯一一家 AI 3D 公司，与 OpenAI、Anthropic、Canva、Superhuman、Higgsfield 等企业共同上榜。
Business Standard利空AI 评分3030 AI a 'classic bubble' that is nearing bursting point, warns Ray Dalio
His comments come as technology giants spend hundreds of billions of dollars on AI, increasingly financed with debt, and as market gains concentrate in a handful of stocks
卫报AI 评分1616 Who owns ‘brain rot’? Inside the international legal drama over a beloved meme
Copyright law in the US only applies to human-made creations, but AI characters like Tung Tung Tung Sahur are challenging thatOne of the most beloved characters in an absurdist genre of online video content has become the center of an international legal battle that could determine where human creat
钛媒体AI 评分5353 OpenRouter与两类大模型评测榜单的口径和排名可比性分析
文章分析OpenRouter调用量榜、Artificial Analysis评测和ARC-AGI榜单，指出三者的统计范围、评分权重、接口规则及版本更新都会影响排名的解释。
钛媒体AI 评分7272 郭台铭旗下工业富联上半年营收5578.6亿元、净利润约237亿元，AI服务器业务增长
工业富联2026年上半年营收5578.6亿元，同比增长54.6%；净利润约237亿元，同比增长95.99%，首次突破200亿元。文章梳理鸿海系上市公司的业务布局，并列出工业富联2025年云计算板块营收6026.79亿元、同比增长88.70%，以及AI服务器业务相关数据。
IT之家AI 评分3939 “AI 教父”辛顿提议 AI 行业建立 FDA 式审批机制：模型推出前需通过安全审核
IT之家 10 月 7 日消息，杰弗里 · 辛顿（Geoffrey Hinton）希望 AI 企业跨过一道制药行业早已熟悉的监管门槛。在当地时间周二播出的《Smart Girl Dumb Questions》播客节目当中，这位被称作“AI 教父”的诺贝尔奖得主、计算机科学家提出建议：AI 公司在产品正式发布之前，必须先向监管机构证明自身产品具备安全性。“你不能直接研制一款新药就把它推向市场。”辛顿对播客主持人纳伊玛 · 拉扎（Nayeema Raza）说道，“你必须说服美国食品药品监督管理局（FDA）。而为了完成这件事，你需要开展大量工作，相关投入大约要十亿美元级别。”“对于人工智能而言，至少
Business Insider利空AI 评分4949 摩根大通CEO杰米·戴蒙称 Anthropic 的 Mythos 大幅加剧网络安全风险
摩根大通CEO杰米·戴蒙称，Anthropic 的 Mythos 使网络安全风险上升了10倍。戴蒙表示，AI暴露出企业此前未知的漏洞；他主张通过切实的安全措施应对风险，而非陷入恐慌。
IT之家AI 评分3434 微软 Surface Laptop Ultra 笔记本规格曝光：最高 128GB 内存，面向 AI 开发和内容创作
IT之家 10 月 7 日消息，科技媒体 Windows Central 昨日（10 月 6 日）发布博文，分享了微软 Surface Laptop Ultra 笔记本的 2 种规格配置，包含 18 核和 20 核心 RTX Spark 方案，目前尚不清楚具体售价信息。IT之家援引博文内容，附上相关规格信息如下：项目规格平台／芯片NVIDIA RTX Spark Superchip，集成 NVIDIA Grace CPU 与 Blackwell RTX GPU，原生支持 CUDA。处理器选项18 核 CPU，搭配 5120 核 GPU；20 核 CPU，搭配 6144 核 GPU。统一内存（R
Business Insider利空AI 评分3838 The Godfather of AI wants an FDA-style approval system for the technology
Geoffrey Hinton said AI companies should have to convince a regulator their models are safe before releasing them, much like drugmakers.
钛媒体AI 评分4141 企业级 AI 订单为何不只拼模型：Palantir 的驻场工程打法正被 OpenAI 和 Anthropic 借鉴
Palantir 通过前线部署工程师驻场、在客户数据和流程中交付系统，建立企业级 AI 项目所需的信任与落地能力。OpenAI 设立部署公司并收购 Tomoro，Anthropic 与 DXC 合作培训前线部署工程师，也在招聘驻场岗位。麻省理工 NANDA 项目报告测算，95% 的企业生成式 AI 试点未产生可量化业务收益，问题大多出在部署环节。
极客公园AI 评分3434 派拉蒙 1100 亿美元收购时代华纳，好莱坞诞生新巨无霸；月之暗面被曝完成 500 亿美元 PreIPO 融资；微软、Meta 被曝要求员工减少使用 Claude
2026 诺贝尔物理学奖揭晓：把一立方公里南极冰变成望远镜，打开「中微子天文学」 10 月 6 日，2026 年诺贝尔物理学奖授予比利时裔美国物理学家 Francis Halzen，以表彰其「对 IceCube 中微子天文台的决定性贡献，以及发现来自天体的高能中微子」。 Halzen 领导建设的 IceCube 位于南极，把超过 一立方公里的透明冰层本身变成粒子探测器：5000 多个光学传感器被埋入超过两公里深的冰层，通过捕捉中微子极少数与物质碰撞时产生的微弱光线，反推出粒子的来源。 与传统望远镜利用光观察宇宙不同，中微子几乎可以不受阻碍地穿过物质，因此能够携带超新星、黑洞等极端宇宙事件的信息
IT之家AI 评分4949 谷歌发布 AI 图像模型 Nano Banana 2.1，提升视觉设计、蒙版编辑与主体一致性
谷歌正式发布 AI 图像生成与编辑模型 Nano Banana 2.1，重点改进视觉设计、蒙版编辑和主体一致性。该模型已逐步部署至 Gemini 应用、谷歌搜索 AI 模式等平台，并向开发者开放，模型 ID 为 gemini‑nano‑banana‑2.1，可输出 1K、2K 和 4K 分辨率图像。
路透社利好AI 评分4545 路透社：英国《金融时报》称SpaceX拟投入400亿美元采购英伟达芯片
英国《金融时报》报道称，SpaceX拟投入400亿美元采购英伟达芯片。
IT之家AI 评分5252 OpenAI 发布未公开前沿模型解决数学难题的研究成果
OpenAI 公布一批由未发布前沿模型解决的数学研究成果，相关文件包含 722 份手稿，涵盖 372 个结果家族及数百个未解决问题。发布资料还包括部分推理过程摘要、算力估算和尝试求解的问题数量统计；OpenAI 称，普通成果的算力消耗约相当于 ChatGPT Pro 连续思考三小时。
The Verge利好AI 评分1313 OpenAI drops another batch of mathematical breakthroughs
OpenAI has revealed solutions to a number of long-standing mathematics problems produced by an unreleased frontier model in a batch of 722 manuscripts, covering 372 result families that group related papers. It extends a run of breakthroughs that have both impressed and unsettled parts of the mathem
金融时报AI 评分6060 SpaceX拟融资400亿美元购买英伟达芯片，Apollo牵头融资
SpaceX拟融资400亿美元用于购买英伟达芯片，融资由Apollo牵头。金融时报称，这笔债务融资显示了支撑AI发展的芯片及其他基础设施投入规模。
IT之家AI 评分6464 谷歌推出多模态嵌入模型 EmbeddingGemma 2，量化后纯文本权重约需 191 MB 内存
谷歌推出 EmbeddingGemma 2，将嵌入能力从文本扩展到代码、图像、视频和音频，并把这些内容映射到同一嵌入空间。该模型总参数为 7.4 亿，支持 8K Token 上下文；量化后在谷歌 Pixel 11 Pro 上，纯文本权重约需 191 MB 运行内存，完整多模态模型约需 567 MB。
IT之家AI 评分6060 Anthropic 扩大网络安全模型测试计划，分层开放 Claude 模型
Anthropic 公布新版网络核验计划（CVP），将此前运行的两项计划整合为三个层级，向通过审核的网络安全人员开放较少安全防护限制下的 Claude 模型测试。
Ars TechnicaAI 评分4444 OpenAI将在欧盟默认对ChatGPT生成文本添加水印
OpenAI宣布将在欧盟自动为ChatGPT生成的文本添加水印，以遵守《欧盟人工智能法案》的标记要求。其他地区也可使用该功能，但默认关闭；OpenAI称其专有方法为textGrain，并将向有限数量的研究人员和组织开放检测工具，其他机构可申请加入。
卫报利空AI 评分2424 Is Australia ready for a datacentre boom? – podcast
As technology companies push to expand AI infrastructure across Australia, residents from suburban Melbourne to regional Queensland are raising concerns over noise, power demands and local impacts. Could growing community opposition slow down plans for a nationwide infrastructure boom?Technology rep
卫报利空AI 评分1717 OpenAI admits misstep in handling AI agent interactions with Australian government sites – video
In a parliamentary inquiry, OpenAI chief strategy officer Jason Kwon is questioned by independent senator David Pocock about the company's notification process after OpenAI agents accessed Australian government website data in June. Asked why OpenAI relied on an arbitrary department email rather tha
卫报利空AI 评分2020 Stop worrying about my AI, says multibillionaire Sam Altman. So we bear the risks and he keeps the money | Chris Stokel-Walker
OpenAI’s CEO seems to want to privatise his company’s huge financial gains while making all of us pay for any negative consequencesTech bros’ messianic streaks can show up in subtle ways. But one of Sam Altman’s flaws is that he often lacks that subtlety. “We believe that the world should accept som
The Verge利好AI 评分5757 Google与Constellation签署20年协议，升级美国六处核电站以供电数据中心
Google与Constellation签署一项为期20年的协议，将升级美国六处核电站，为数据中心增加电力供应。根据新闻稿，该购电协议旨在保障收入，以覆盖现有核电站效率升级成本，Constellation称这代表超过43亿美元的新投资；双方预计未来五年各站新增设备合计带来890兆瓦新增装机容量。
TechCrunch利好AI 评分3333 How AI decision models could change content moderation
On Tuesday, Musubi announced a lightweight decision model made for real-time moderation called PolicyLM-1.7B, released with open weights.
TechCrunch利好AI 评分3030 Silicon Valley’s AI wunderkind launches Underdog, the most private Instinct/Muse competitor yet
Sigil Wen, backed by a Silicon Valley's who-who, has built an on-device AI assistant that promises to be free, fully private, and capable for everyday tasks.
财富AI 评分4444 OpenAI人权负责人Sarah Yager谈军事如何使用AI：令她夜不能寐
OpenAI人权与负责任部署负责人Sarah Yager担心军事机构在武装冲突中使用AI，尤其是用于目标识别和自主武器。她称自己正与研究人员、工程师合作改进模型对战争法等国际协议的回应，也在帮助政府监督机构建立追踪政府AI使用情况的工具。
路透社利好AI 评分3636 Anthropic opens its most powerful AI models to more security teams - Reuters
Anthropic opens its most powerful AI models to more security teams Reuters
TechCrunchAI 评分1919 The next hurdle for AI agents: getting websites to let them in
Personal AI agents promise to shop, book flights, and make reservations for you. But deliberate blocks and anti-bot defenses are getting in the way, leaving consumers caught in the middle. A new standard aims to help.
TechCrunch利好AI 评分5757 AI计算初创公司Lambda拟融资至多40亿美元，计划2027年IPO
AI计算初创公司Lambda拟融资至多40亿美元，融资前估值为145亿美元，并计划于2027年IPO。此次融资由Coatue和Blackstone领投，Lambda获得英伟达支持。
路透社利好AI 评分4040 路透社：Nvidia-backed Lambda 拟在计划中的 IPO 前融资 40 亿美元，WSJ 报道
WSJ 报道，Nvidia-backed Lambda 计划在拟议中的 IPO 前融资 40 亿美元。
WiredAI 评分1212 ‘Artificial’ Roasts AI’s Creators—and Sends a Dark Warning About Its Dangers
From their dorky parties to their weird walks, the movie holds OpenAI CEO Sam Altman and other stakeholders with contempt, while demonstrating their recklessness.
CNBC利好AI 评分66 What Marvell's rosy long-term guidance means for our AI chip stocks
Every weekday, the Investing Club releases the Homestretch; an actionable afternoon update just in time for the last hour of trading.
WiredAI 评分1515 OpenAI Is Pissing Off a Bunch of Mathematicians—Again
“There’s a perception of mobster behavior” from leading AI companies, one mathematician tells WIRED as OpenAI prepares to release more than 100 new solutions to unsolved problems.
Business InsiderAI 评分2626 The Anthropic researcher whose resignation made waves last month says he was 'working to automate' himself
Jacob Coxon, who resigned from Anthropic last month, warned that once AI can conduct its own research, "humans will be much further out of the loop."
彭博利空AI 评分1212 Jamie Dimon Warns on AI's Growing Cyber Threat
Get a jump start on the US trading day with Dani Burger on "Bloomberg Open Interest." Stocks push toward new highs, but Ray Dalio is sounding the alarm on bonds. JPMorganChase Chairman & CEO Jamie Dimon joins us exclusively on rising debt, dealmaking and AI’s growing cyber threat. Plus, Nvidia nears
Fast Company利好AI 评分3838 How Hark designed an AI assistant that’s actually helpful
With a chatbot interface and agentic capabilities, the new AI web application Hark Pro checks the main boxes of a mass market AI release in 2026. But unlike the tools that have already been released by the large tech companies and frontier AI labs, Hark Pro is rethinking the core premise behind the
The VergeAI 评分1616 We can’t just change the definition of ‘recording’
With AI hardware, tech companies are pushing the definition of what does and doesn't constitute a recording. For most of gadget history, it'd be reasonable to assume that a device with a microphone or camera is either recording you or it isn't; it's either on or off, without much gray area. Micropho
TechCrunch利好AI 评分1414 Mirror Particle is building a ‘world model’ of human behavior
Mirror Particle will launch at TechCrunch Disrupt's Startup Battlefield 200 with a world model built from scratch to predict human behavior, arguing that LLM role-play falls short for market research and brand strategy.