资讯时间轴
1562 条 · 11 个来源正文已在本站存档，原文失效也不会变成空页；已剔除失效外链与不合规内容。
Apple CEO says the company doesn’t ‘want to ship AI for AI’s sake’
Apple's new CEO defended the company's current position in the AI landscape.
来源：TechCrunch AI · 本地全文 站内详情4 days left: How TechCrunch Founder Summit helps you make more useful connections to scale
On November 4 at Boston’s SoWa Power Station, get ready for a full day packed with meaningful networking opportunities, actionable insights, and connections tha…
来源：TechCrunch AI · 本地全文 站内详情FilingLoop: Evaluating and Evolving LLM Agents for Corporate Disclosure Review and Revision
Public disclosure repositories such as SEC EDGAR support informed investment through web access to corporate filings. Supporting their review with language mode…
来源：arXiv cs.AI · 本地全文 站内详情Disparity Laundering: Consensus Without Fairness in Clinical Multi-Agent Systems
Clinical large language models are increasingly fielded as multi-agent teams, on the assumption that deliberation averages out individual error. It does not ave…
来源：arXiv cs.AI · 本地全文 站内详情SCOPE: Selective Critical-Step On-Policy Self-Distillation for Agentic Reinforcement Learning
Outcome-based reinforcement learning provides coarse supervision for long-horizon LLM agents: failed trajectories may contain correct steps, while successful tr…
来源：arXiv cs.AI · 本地全文 站内详情Agent Transfer Protocol: Server-Mediated Agent-to-Agent Communication Architecture for the Agentic Web
The Agentic Web needs to support continuous collaboration among agents from different organizations. These collaborative relationships may span multiple tasks, …
来源：arXiv cs.AI · 本地全文 站内详情DES-JEPA: Learning a Shared Vocabulary for High-Dimensional Discrete Event Sequences
Temporal foundation models such as Chronos and TimesFM generalize across domains through a shared numerical observation space. In contrast, discrete event seque…
来源：arXiv cs.AI · 本地全文 站内详情Accelerating Nonlinear MPC by Behavior Cloning with Separated Timescales: Application to Nuclear Reactor Load-Following
Behavior Cloning (BC) of expert controllers for multiscale dynamical systems is challenging: state variables evolving at widely separated timescales induce high…
来源：arXiv cs.AI · 本地全文 站内详情A-RSI: Adaptive Task Refresh for Zero-Data Multimodal Self-Improvement
Recursive self-improvement requires models not only to acquire useful training experience, but also to decide when their current experience has ceased to be use…
来源：arXiv cs.AI · 本地全文 站内详情阿里巴巴与 Apple 达成合作，将千问集成至 Apple 智能
10月31日，阿里巴巴与Apple达成合作，将千问集成至 Apple 智能，中国用户可以在Siri、写作工具及视觉智能中，直接调用千问的文本与图像理解、内容生成等能力。这些功能将于11月4日起随最新的iOS 开发者 Beta 版开放测试，并将于12月iOS 27.3正式发布时面向中国用户推出。 Siri可以直接调用千问…
来源：雷锋网 · 本地全文 站内详情Can AI data centers be mass-produced like cars? This startup aims to find out
Runware is building a factory to turn out prefabricated AI data centers on an assembly line. The real test will be whether it can deliver at that scale.
来源：TechCrunch AI · 本地全文 站内详情李飞飞挖的三个「坑」：从 ImageNet 到空间智能，这一次她会怎么填？
82 亿美元，已经为整个世界模型赛道提供了价值锚点。 作者丨幸丽娟 编辑丨岑 峰 82亿美元，全股票收购李飞飞创立的 World Labs。这是 AMD 史上第二大收购，仅次于 2022 年以约 500 亿美元拿下赛灵思。赛灵思是30多年的老牌芯片公司，而 World Labs，仅仅成立两年。这一消息发出后，圈内人士的…
来源：雷锋网 · 本地全文 站内详情聚力开源向实 赋能产业创新——2026 上海开源软件应用创新大赛总决赛暨创新增长大会圆满落幕
10月29日至30日，2026上海开源软件应用创新大赛总决赛暨创新增长大会在上海张江人工智能创新小镇举行。本届大赛由开源中国主办，自8月5日启动报名以来，共收到报名项目766个，40余家合作企业和开源社区参与；12道真实产业命题吸引269个项目应题攻关，30支团队晋级总决赛，角逐总价值100万元的奖池。10月30日的颁…
来源：OSC 开源中国 站内详情盖伦奖提名一年后，希格生科的边界、验证与野心
“五年前我们把这页PPT放出来，很多人都说这是在‘画饼’。但现在我们和晶泰合作研发的全球首款弥漫性胃癌靶向药，已迈向关键临床验证。” 作者丨周蕾 编辑丨岑峰 【编者按】被誉为"医药界诺贝尔奖"的美国盖伦奖（Prix Galien USA），于10月29日公布了2026年度获奖名单。 值得注意的是，晶泰科技孵化的创新企业…
来源：雷锋网 · 本地全文 站内详情在代谢酶网络里，默达生物找到了一个让坏细胞「吃不饱饭」的开关
“相较于复杂的生物学预测，我们目前更看重AI在海量筛选方面的应用价值。” 作者丨周蕾 编辑丨岑峰 【编者按】被誉为"医药界诺贝尔奖"的美国盖伦奖（Prix Galien USA），于10月29日公布了2026年度获奖名单。 值得注意的是，晶泰科技孵化的创新企业已连续两年入围该奖项提名。去年，希格生科凭借全球首款弥漫性胃…
来源：雷锋网 · 本地全文 站内详情伯牙智能刘欣：GPT-6 之后，轮到硬件拖机器人后腿了 | 物理AI 50人
灵巧手去年出货近 2 万只，大部分在「吃灰」 作者丨向 欣 编辑丨高景辉 去年，国内灵巧手公开可查的出货量是约两万只。在伯牙智能创始人刘欣的观察里，其中大部分灵巧手的归宿是跟着人形机器人表演几场，然后被搁在角落里，「表演完就吃灰」。便宜的低自由度灵巧手多用于表演；操作能力更强的高自由度产品又容易损坏，有的因为贵，买回来…
来源：雷锋网 · 本地全文 站内详情解读丨Misa 的全彩豪赌：卖不到千万台，Rokid 就输了
提词器是原生应用，但依然不是杀手级应用。作者丨黄艳冰 编辑丨姚单 “眼镜的 iPhone 时刻来了。” 10月27日，Rokid 创始人兼 CEO Misa 在首款全彩 AI 眼镜 Rokid Color 亮相后，抛出了这句话。 与市面上已有的全彩尝试相比，Rokid Color 讲的并非新故事，不同之处在于，它在 5…
来源：雷锋网 · 本地全文 站内详情脑景中心启幕、四款新品齐发！熵基科技加码非侵入式脑机接口全栈技术与应用平台
10月29日，以"BE THE CHANGE MAKER"为主题的熵基科技全球合作伙伴大会2026在东莞启幕，会期将持续至10月31日。本届大会聚焦原生AI发展趋势，围绕空间智能、脑机接口与人工智能融合创新，以及全球产业生态协同展开，集中呈现熵基科技在技术创新、产品研发与场景应用方面的最新进展。 大会期间，ZKBrai…
来源：雷锋网 · 本地全文 站内详情卓驭发布“磐行”智驾解决方案：切入商用车市场，一年可回本
10月29日晚，智能驾驶方案商卓驭科技正式发布商用车智能辅助驾驶解决方案“卓驭磐行”，并公布了商用重卡、客车、轻卡、ROBOVAN等业务进展。 此次卓驭磐行分为PDS 4和PDS 5两个版本。 PDS 4对应高悟性大模型4.0及以上版本，主要支持高速NOA，面向重卡干线物流。目前这一版本已经实现量产。 PDS 5则对应…
来源：雷锋网 · 本地全文 站内详情AWS will be a $1T business, Jassy says. But first, it will spend big.
Amazon wants to be the king of AI cloud computing and become the first $1 trillion revenue business. But to do that it will have to continue on its capex spendi…
来源：TechCrunch AI · 本地全文 站内详情Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity
Mistral says its new model can afford organizations a kind of control over its cyber defenses that closed models can't.
来源：TechCrunch AI · 本地全文 站内详情U.S. investigated Chinese spy plot aimed at Nvidia CEO
A top executive of a Singapore AI cloud company that was accused of illegally selling Nvidia chips to China could also have been a Chinese spy, sources have tol…
来源：TechCrunch AI · 本地全文 站内详情Deepgram acquires Pipecat-maker Daily to bolster its voice AI infrastructure
Deepgram said that it is committed to keep Pipecat open source.
来源：TechCrunch AI · 本地全文 站内详情AI裁员背后：消失的不只是工作，还有年轻人的机会
企业用AI省下的人工成本已经写进财报，但裁员公告远远装不下这场变化。那些没有被招聘的人、逐渐消失的职业入口，以及技术收益的重新分配，正在成为更大的问题。
来源：钛媒体 · 本地全文 站内详情Wiz’s CTO on PHP2Shell and the coming ‘vulnpocalypse’
The PHP scare may have faded from your feed, but the cleanup is far from over. Last week, two critical bugs in the open source language behind roughly 70% of we…
来源：TechCrunch AI · 本地全文 站内详情The ‘Vulnpocalypse’ is coming for the internet’s aging code, says Wiz’s CTO
The PHP scare may have faded from your feed, but the cleanup is far from over. Last week, two critical bugs in the open source language behind roughly 70% of we…
来源：TechCrunch AI · 本地全文 站内详情OpenAI and ServiceNow expand partnership across enterprise workflows
OpenAI and ServiceNow expand their partnership to bring OpenAI models and agents into ServiceNow and ServiceNow workflows into ChatGPT and Codex.
来源：OpenAI 站内详情字节辟谣"群演数据喂大模型"：严重失实！
近日，有网络传言称"横店与字节合作，将15万名群演的肖像和声音喂给字节大模型"。字节豆包公关负责人刘星今日通过微博作出回应，表示该传言严重失实。 据介绍，此次引发讨论的项目为横店影视AI服务平台建设项目。刘星表示，该项目的初衷是希望建设一套影视行业的公共数字基础设施，把影视基地实景等资产做成高质量数据集，帮助剧组更高效…
来源：雷锋网 · 本地全文 站内详情From Solver Trajectories to Teaching Trajectories: Cognition-Aligned Reasoning Distillation
Knowledge distillation from large reasoning models (LRMs) often uses raw teacher trajectories as supervision. However, this practice creates a mismatch between …
来源：arXiv cs.AI · 本地全文 站内详情IBM CEO 力挺开源权重模型，称其在本地部署方面极具吸引力、且价格低廉
IBM CEO Arvind Krishna 在 Q3 财报电话会上算了一笔账。他说开源权重模型现在能做到前沿闭源模型 90% 的效果，只花 10% 到 25% 的 token 成本。"这个问题对企业来说很简单：你想要一个只要 10% 价格、能达到最好模型 90% 水平的东西，还是愿意付十倍的钱？" 他自己回答了：绝大…
来源：OSC 开源中国 站内详情Parents sue OpenAI, alleging GPT-4o coached their daughter off seizure medication
A family is suing OpenAI, saying ChatGPT coached their daughter to quietly taper off her seizure medication and keep it from her doctors before she died in Febr…
来源：TechCrunch AI · 本地全文 站内详情Meta AI director Rohit Patel leaves to take on Microsoft Office and Google Workspace
Rohit Patel is departing Meta after eight years working on AI evaluations and reinforcement learning, TechCrunch has exclusively learned. His startup, Ndoc, aim…
来源：TechCrunch AI · 本地全文 站内详情千问办公接入乐奇AI眼镜，一句话发起任务
10月27日，在乐奇AI眼镜新品发布会上，乐奇Rokid和千问办公宣布达成合作，千问办公正式接入乐奇AI眼镜。用户在“Rokid AI”App里的智能体商店添加“千问办公”，绑定千问办公账号后，就可以戴上眼镜用语音唤起千问办公、下发任务，并在眼镜端查看结果。 比如，用户直接说：“千问办公，帮我调研一下AI眼镜的行业现状…
来源：雷锋网 · 本地全文 站内详情OpenAI expands partnership with Oracle to bring its apps to ChatGPT
OpenAI and Oracle are expanding their partnership to bring business applications and AI agents into ChatGPT and power new Oracle offerings.
来源：OpenAI 站内详情Inverse Modeling for Laser Pulse Shape Design in Inertial Confinement Fusion
The achievement of practical fusion energy remains one of the most pressing unsolved scientific challenges, with immense implications for carbon-free power. A c…
来源：arXiv cs.AI · 本地全文 站内详情TAD: Distilling Vision-Language-Action Policies from Teacher Trajectories Alone
Online distillation of vision-language-action (VLA) policies commonly relies on teacher action scores that an inference-only interface may not expose. Trajector…
来源：arXiv cs.AI · 本地全文 站内详情Local Entropy Fluctuations for Black-Box LLM-Generated Text Detection
Zero-shot detection of machine-generated text is vital for mitigating the risks of large language models (LLMs) misuse. While the state-of-the-art dual-model de…
来源：arXiv cs.AI · 本地全文 站内详情新华投控战略投资九识：赋能智能驾驶与物理AI，共筑智慧物流“中国路”
10月23日，九识与新华社投资控股有限公司（新华投控）签署战略投资协议，达成深度战略合作。这是新华投控在智能驾驶和物理AI领域的首个产业直投项目。双方将在品牌塑造、智慧物流生态构建及海外拓展三方面强强联合。九识今年实现跨越式发展：成为行业首个百亿独角兽，实现全球首个L4级无图方案规模化量产，除车辆销售收入外，AI Dr…
来源：雷锋网 · 本地全文 站内详情知名具身智能公司陷欠薪停工风波：员工称欠薪近800万元，CEO称联系不上实控人；iPhone Duo被曝屏幕亮线开裂！苹果客服回应：非硬件问题
要闻提示 1.知名具身智能公司陷欠薪停工风波：员工称欠薪近800万元，CEO称联系不上实控人 2.iPhone Duo被曝屏幕亮线开裂！苹果客服回应：非硬件问题 3.00后美女经理玩梗“老登”遭吐槽：小鹏总裁悬赏100万澄清 4.宇树机器人侧空翻表演甩断腿，前一天刚演示「踢凹车门」 5.字节商业化GenAI负责人文石磊…
来源：雷锋网 · 本地全文 站内详情端到端推理时延降低 15%：京东广告联合鲲鹏超节点，探索生成式广告行业新路径
京东广告面临生成式推荐带来的千倍算力需求与300毫秒时延红线之间的尖锐矛盾，真正的瓶颈在通算侧而非GPU。为此，京东与华为鲲鹏合作，通过灵衢互联将四级缓存压成两级，建成300TB共享内存池，并重构RPC通信，使端到端推理时延降低15%，KV Cache和RPC性能大幅提升，经618大促核心场景5%流量验证。双方合作正从…
来源：钛媒体 · 本地全文 站内详情Tradeoff-Aware Territory Design: A Multi-Objective Decision System for Enterprise Customer-to-Seller Allocation
Enterprise customer journeys are shaped not only by product experiences and marketing, but also by recurring resource-allocation decisions. In B2B go-to-market …
来源：arXiv cs.AI · 本地全文 站内详情Knowing Whom to Trust is not Enough: The Trust--Agreement Gap in Repeated Multi-Party LLM Negotiation
Large language model (LLM) agents increasingly participate in negotiations on behalf of individuals and organizations, yet they remain vulnerable to manipulatio…
来源：arXiv cs.AI · 本地全文 站内详情Behaviour analysis for large language model-driven agents
Understanding language-agent behaviour requires learning how actions are organised across a trajectory and how this organisation relates to context and outcomes…
来源：arXiv cs.AI · 本地全文 站内详情Correct Answers, Closed Gates: SolveBind for GUI Agents at the Verification Barrier
Websites and applications deploy CAPTCHAs and anti-automation mechanisms that use behavioral signals such as interaction trajectories and timing, not outcomes a…
来源：arXiv cs.AI · 本地全文 站内详情Financial Context-Rich Strategy for Actor-Critic based Agents: A Deep Reinforcement Learning Approach for Stock Trading
In modern financial markets, more than 3/4th of the large volume trade is executed by algorithmic trading based on advanced trading strategies to attain higher …
来源：arXiv cs.AI · 本地全文 站内详情When to Stop Penalizing Actions? Martingale GRPO for VLA Reinforcement Learning
A failed vision-language-action (VLA) rollout does not imply that every action contributed to failure. Yet in group-relative policy optimization (GRPO), a singl…
来源：arXiv cs.AI · 本地全文 站内详情Edge AI Daily 早报（10月26日）
苹果终结二十年触控抵制，首款触屏OLED MacBook Pro定档发布，成为特努斯接任CEO后首个硬件答卷；Meta自由现金流转为-21亿美元，7600亿美元行业资本支出迎盈亏平衡审判；谷歌以1070兆瓦虚拟电厂重构数据中心供能，特斯拉自研AI地图弃用谷歌；Anthropic被指静默削弱Claude Opus 5.5…
来源：钛媒体 · 本地全文 站内详情“更懂你、会办事”，支付宝携手科大讯飞升级智能座舱服务
10月21日，支付宝车载助手亮相科大讯飞智能汽车产品技术发布会，宣布与讯飞星火智慧座舱战略合作升级，正式启动「讯宝计划」，推动智能座舱从信息交互向生活服务延伸，让车主在车内即可办理更多日常事物。 四大生活服务场景将落地，覆盖出行高频需求 发布会上，支付宝车载助手接入星火智慧座舱3.0的合作成果正式亮相。依托讯飞汽车 M…
来源：雷锋网 · 本地全文 站内详情iRTE2026：对话 AI 如何迈向 AGI？声网CEO赵斌给出五级对话能力框架
10月23日，iRTE2026实时智能大会在北京举行。声网创始人兼 CEO 赵斌在主论坛发表题为《对话 AGI：实时互动的下一章》的演讲，回顾了过去一年实时互动、对话式 AI 与底层基础设施的关键变化，并提出“五级对话能力”框架，勾勒对话式 AI 从基础应答走向深度理解与关系沟通的演进路径。 赵斌认为，随着 AI 从回…
来源：雷锋网 · 本地全文 站内详情车企造人形机器人，究竟图什么？
车企入局人形机器人，依托汽车产业资金、制造与线下网络，先在工厂、门店落地测试迭代。各家路线分化：自研、合作验证、渠道联动并存。短期优化生产与门店体验，长期谋求第二增长曲线，但机器人仍需跨过可靠性、成本门槛，商业化前景尚不确定。
来源：钛媒体 · 本地全文 站内详情早报｜微信初步适配iPhone Duo，仍不能平板登录/全球PC出货下降21%/何刚出任华为董事
黄仁勋：对长鑫合作持开放态度，不认为 2029 年内存会过剩 · 沃尔沃汽车拟全球裁减约 2500 个岗位 · 华为 Pura90 Plus 开启预订#欢迎关注爱范儿官方微信公众号：爱范儿（微信号：ifanr），更多精彩内容第一时间为您奉上。
来源：爱范儿 · 本地全文 站内详情Hottest toy for Christmas? Nvidia CEO plays with adorable Microduck robots, 20,000 sell
The robots, stomping their little feet and making tiny R2-D2 noises, sat, fell down dead for a "nap," and summersaulted on voice command.
来源：TechCrunch AI · 本地全文 站内详情From Toast to Bessemer: Who founders will hear from at TechCrunch Founder Summit 2026
Hosted by TechCrunch in partnership with Fidelity Private Shares, the full-day event goes beyond fundraising and AI to explore the realities of finding customer…
来源：TechCrunch AI · 本地全文 站内详情Europe Is Sorting China’s Car Brands by Showroom, Not by Factory
Chinese automakers can build competitive vehicles at home. In France and across Europe, results now split on which models arrive first, how dealers get paid, an…
来源：钛媒体 · 本地全文 站内详情Hidden Messages Between Diffusion Language Model Agents: Causal Evidence for Latent Communication
Diffusion language models form answers through a sequence of revisable continuous states, but multi-agent systems usually discard those states and make agents c…
来源：arXiv cs.AI · 本地全文 站内详情Robust Soft Skill Adaptation via Behavioral Localization in Tool-Using Agents
Text skills can extend the capabilities of tool-using agents, but long skill descriptions increase computational costs. Soft prompts provide a compact alternati…
来源：arXiv cs.AI · 本地全文 站内详情72岁董明珠：被学生喊奶奶很不开心，人的价值不用年龄来衡量；雷军：汽车产业没有安全，其他都是零；iPhone Duo全球限购，每人限购2台
要闻提示 1.iPhone Duo开启全球限购，每位顾客仅可购买2台2.72岁董明珠：被学生喊奶奶很不开心，人的价值不用年龄来衡量3.雷军：汽车产业没有安全，其他都是零4.星巴克回应消费入座争议：10月22日起全国施行并摆放提示牌5.宇树发新人形机器人！能徒手抬汽车，还会勇斗歹徒6.离职Kimi后，17岁天才少年陈广宇…
来源：雷锋网 · 本地全文 站内详情Microsoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’
Microsoft’s new AI diagnostic system beat generalist doctors on hard cases, according to a study in Science. But most of the improvement seems to come from the …
来源：TechCrunch AI · 本地全文 站内详情Spotify’s current AI bet is about being the music service that agents choose
Spotify co-CEO Gustav Söderström explains why the rise of AI agents could be good news for the music service.
来源：TechCrunch AI · 本地全文 站内详情Flowtica Rian：硬件的旧规律没变，新机会是做好Memory丨未来创新50人
大疆想收编的云台团队，十年做了四个硬件品类，现在押注 AI Memory。作者丨吴优 编辑丨董子博 Plaud 爆火之后，国内一家叫 Flowtica 的公司也在短时间内闯入大众视野，去年上半年在 Kickstarter 平台上发布了世界首款 AI 笔 Flowtica Scribe，最终筹得 33 万美金。这支笔仅重…
来源：雷锋网 · 本地全文 站内详情从“卖分子”到“卖平台”：这单BD改了中国AI制药的站位
这既是礼来首次押注AI驱动的脂质纳米颗粒技术，也是国内首个生成式AI-LNP出海大单、全球规模领先的肝外靶向LNP平台合作。礼来采购的，是一套能够持续产出新技术方案的系统能力。
来源：钛媒体 · 本地全文 站内详情OpenAI acquires Orchestra Bio
OpenAI is acquiring Orchestra Bio to bring AI-powered operations to life sciences R&D, helping biopharma teams accelerate drug development and advance new treat…
来源：OpenAI 站内详情How Itaú is building generative AI for its entire customer base
With ia.i, a conversational interface, and the IARA governance platform, Itaú is applying its AI First strategy to customers and employees — in one of Brazil’s …
来源：OpenAI 站内详情