Skip to content

All news

54 today

Oct 28Wednesday

The Verge · AI

Google’s next Genie AI model generates more realistic experiences

Google 今天宣布世界模型 Genie 4,可生成更逼真的交互体验,支持用自然语言提示词实时修改体验并生成音频。模型使用 Gemini 3.8 Flash 辅助模拟,体验时长从 Genie 3 的一分钟增至约两分钟,Google 提醒较长模拟中的细节和物体持续性可能发生变化。

TechCrunch · AI

Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering

Google DeepMind 于周二发布 Genie 4 研究预览,新增实时音频与模拟中的提示词编辑,生成更逼真、可交互的世界。模型可实时改变天气、镜头角度或添加人物与物体,保持场景一致性最长两分钟,是前代的两倍,但仍存在细节漂移、音频准确性和场景文字可读性问题。

Google DeepMind

Google DeepMind releases Genie 4 world model with native audio and real-time interactive simulation

Google DeepMind released Genie 4, a world model with native audio, interactive dialogue and real-time scene edits in natural language. It is in research preview for trusted testers. In simulations, Gemini 3.8 Flash keeps observing the scene, breaking down goals and checking progress to hold multi-step events together. The model still drifts on object persistence, complex physical interactions and long-term consistency.

Why it matters: Genie 4 puts native audio and live scene control into world simulation, so creators can adjust camera, dialogue and action to try out ideas in a moving scene.

Hacker News front page

New 14‑inch MacBook Pro with M6

Apple 发布搭载 M6 的新款 14-inch MacBook Pro,称其端侧 AI 工作负载性能较 M5 最高提升 25 percent。M6 配备 12-core CPU、12-core GPU 和 Dual 16-core Neural Engine,本地 LLM 提示词处理的首 token 性能较 M1 最高提升 13x、较 M5 最高提升 1.3x。

Computing Life · Share · Yage

DeepSeek 开源 DeepInfer:单模型推理引擎为何常比通用引擎快

DeepSeek 开源的 DeepInfer 仅支持 DeepSeek-V4.1-Flash 离线批量推理,文章解析其性能优势来源与开源范围。GPUStack 同机测试中,DeepInfer 总吞吐比 vLLM 高 25.2%,首字延迟却为 4.28 秒对 2.04 秒,双方均关闭投机解码且未充分调优。

Hacker News front page

Show HN: Specification First Agentic Development Plugin and Prove

spec-first 提供规范优先的 AI 智能体开发插件,每项变更先审查规范,再在独立分支和 worktree 中实现,通过项目检查后才能提交。流程涵盖 create-spec、review-spec、apply-spec 和 deliver-spec,并将决策记录与恢复状态保存在文件中,支持将一系列规范并行构建后合入一个 pull request。

Hacker News front page

Petroly developer builds a 15-second promo video with unit tests using Claude Code

A Petroly developer used Claude Code to write a JavaScript renderer that produced a 15-second promo video with unit tests, without opening a video editor. The pipeline generates 375 frames through SVG, sharp and FFmpeg. Tests check that the notification appears once the price hits the target, the list ordering and frame glitches; humans still review the visuals.

Why it matters: It turns a promo video into testable render code, folding animation timing checks and multi-language versions into the same software workflow.

Hacker News front page

Anthropic trained Claude on illegally obtained Stanford professors' books

Anthropic 使用非法获取的逾 10 名斯坦福教授的著作训练 Claude,联邦法官裁定其不得盗版图书以建立永久中央资料库。Bartz v. Anthropic 的 $1.5 billion 和解覆盖近 500,000 部已登记作品,法官同时认定使用合法获取的图书训练模型属于合理使用。

AI HOT · Products

Runway Ads 向所有付费计划开放

Runway Ads 今日向所有付费 Runway 计划开放,将智能体创意工具与广告账户的实际表现数据结合。用户可连接 Meta、Google 和 TikTok,生成符合品牌规范的广告变体、调整语言和尺寸,所有广告均须经团队成员批准后才能发布。Runway 表示,其内部团队使用该工具后,每周广告量从不足 100 增至约 1,000,团队规模不变,每位订阅者获取成本降低了 41%。

Oct 27Tuesday

Google Blog · Gemini

Google shows 8 Gemini 4 Argon use cases, from sorting archives to playable 3D scenes

Google demonstrated eight applications of Gemini 4 Argon, including complex document sorting, interactive tools, playable 3D scenes and animation generation. In Antigravity, the model turned more than 80,000 raw NASA files from 72 Mars flights into a presentation. It also coordinated sub-agents, Blender MCP and Nano Banana to turn a real desk photo into a playable 3D track.

Why it matters: Google shows how Gemini 4 Argon builds a playable 3D scene from a photo, chaining sub-agent coordination and modeling tool calls in one task.

Hacker News front page

Viral Polls Are Not Representative Polls

一项针对 Bluesky 账号 @usapolling.bsky.social 近 90 天帖子的分析发现,越偏离该选区平均民调、对民主党越有利的民调帖子被转发得越多。研究者用 LLM 提取帖子中的民调数据并计算 Dem-Rep 差距,在多数选区都观察到这一模式,仅 NC 参议院选区例外。模型显示决定转发量的是帖子与平均民调的偏离度,而非民调本身的差距。

Latent Space

Coding From the Himalayas With an Offline AI Agent

Rittika Jindal 在笔记本上通过 OpenCode 和 Ollama 运行 Qwen3.5 9B,断网后完成费用计算程序的修复,原有 5 项测试全部通过。教程给出预下载模型与项目依赖、限制为本地服务、检查断网状态及逐次批准编辑和命令的流程,作者也独立重跑测试并确认测试文件未改动。实测仅涉及一行累加逻辑修复,CSV 报告和日志分析器练习尚未完成离线运行,不能据此判断大型仓库任务的表现。

Hacker News front page

Is my data used for model training?

Anthropic 默认不使用 Claude for Work、Anthropic API、Claude Gov 等商业产品的输入或输出训练模型,但用户主动提交反馈、报告问题或授权使用数据时,对话和编码会话可能用于训练。

Hacker News front page

New iMac with M6

苹果推出搭载 M6 芯片的新 iMac,官方称其速度最高达最主流 Intel 版 iMac 的 18 倍、M1 版 iMac 的 12 倍,端侧 AI 性能为 M4 版 iMac 的 5 倍。

Hacker News front page

Here be monsters: What do you still need to know when agents do the work?

AI 智能体承担开发工作后,工程师仍需掌握基础知识、积累知识广度,并能在故障发生时按需深入陌生系统。adrq 以 Codex 生成代码触发 SonarQube 复杂度检查、自己的智能体编排器出现生命周期故障为例,解释识别异常与深入排障的必要性。文章主张建立系统的高层抽象模型,并通过实践训练快速学习和解决未知问题的能力,而非要求逐行理解所有生成代码或完全放弃理解实现。

The Decoder

OpenAI taps outside experts, including a Nobel laureate, to review AI-generated math and physics proofs

OpenAI has invited outside experts, among them Nobel physics laureate David Gross, to review AI-generated proofs in math, computer science and physics. The company says the work includes disproving the exponential time hypothesis, solving a local smoothing conjecture in all dimensions, and establishing the mathematical existence of several four-dimensional quantum field theories. The Yang-Mills result covers existence only, not the mass gap, so it is not a full solution of that Millennium Prize problem.

Why it matters: OpenAI is routing AI research results in three fields through separate review paths, with formal checks and expert review covering logic and meaning respectively.

TechCrunch · AI

Opera introduces a new web standard to let users deploy AI locally in their browser

Opera 推出 WebAI 网络标准,让用户直接在浏览器本地运行 AI 模型,无需依赖云端。该功能已在 Opera One 提供早期预览,需用户主动启用,Opera GX 和 Opera Neon 将随后支持;兼容模型包括 Google 的 Gemma 和 Mac 上的 Apple 端侧基础模型。WebAI 支持本地文本生成、内容分类和信息关联理解,相关数据留在设备上,模型性能取决于设备硬件。

TechCrunch · AI

Japan internet giant GMO says hackers posted stolen customer data

日本互联网公司 GMO Internet 称,其 Onamae 云桌面服务遭未授权访问后,疑似攻击者已公开与客户及业务相关的信息,公司正核实内容真伪,并称 2.7 万台服务器可能受影响。GMO 已建议客户重置密码,此前有客户报告交易账户出现“未授权操作”。黑客曾发邮件索要约 80 亿日元(约 5000 万美元)赎金,公司已向日本警方报案。

Hacker News front page

AI promised less work. I've never worked more

Bold.org 技术负责人发现,AI 智能体降低了实现成本,却让自己承担更多并行任务,审查结果与切换上下文带来了更重的精神负担。截至 2026 年 10 月 27 日,他的 GitHub 贡献达到 2,102 次,超过 2021 至 2025 年任一全年;8 月达到 427 次,为个人记录中最繁忙的月份。

Hacker News front page

New 14‑inch MacBook Pro with M6

Apple 发布搭载 M6 的 14 英寸 MacBook Pro,强化端侧 AI 能力,美国起售价为 $1,999,10 月 28 日开启预订、11 月 6 日发售。

MIT Technology Review · AI

Industrial Intelligence: The next growth opportunity for manufacturing

工业智能将制造商的积累经验、运营实践与生产数据结合,让知识在决策现场可用并接入物理生产环节。Toyota Industries 打通碎片化喷漆车间数据后,分析周期从五天缩短到四小时以内,试点将相关喷漆缺陷减少约 25%。Rockwell Automation 新加坡工厂用维护 copilot 整合制造数据、设备手册与老员工经验,新员工掌握数百台设备排障的时间从九个月降至三个月。

Hacker News front page

Wikipedia's Cleanup of AI-Generated Content

维基百科 WikiProject AI Cleanup 志愿者正在开发工具,识别并移除违反网站规范的 AI 生成内容,排查重点包括细微写作结构和不支持正文主张的引文。

TechCrunch · AI

Nvidia backs Indian startup helping companies run AI models with fewer GPUs

Nvidia 参与印度创企 Simplismart 的 1400 万美元 B 轮融资,后者估值约 1 亿美元,其软件可优化模型部署,让企业用更少 GPU 运行 AI 模型。Simplismart 称已帮 AI 视频平台 InVideo 将 GPU 服务成本降低 56%,客户还包括 HeyGen、Higgsfield 和 Ema。其年化收入运行率过去六个月增长 14 倍,超半数收入来自美国。

Hacker News front page

Roblox downloaded 145GB models from us (including adult content) without asking

Open3DLab披露,一个自称属于Roblox的爬虫从其与SFMLab、SmutBase下载了712个项目的1,027个文件,共145.5 GB。站方称其中345个项目属于成人内容、423个项目仅许可非商业使用,爬虫无视robots.txt与禁止抓取的服务条款,站方封禁后两次邮件询问均未获回复。

The Verge · AI

Data center backlash is uniting America

数据中心快速扩建及其基础设施成本,正成为美国2026年多州国会和州长竞选中跨党派的争议议题。多地民主党与共和党候选人都在收紧立场,但对暂停建设、取消税收优惠、由开发商承担能源和用水成本,以及交由地方社区决策存在分歧。

The Decoder

Nvidia to invest $3 billion in Meta as Meta plans to buy more Nvidia chips

Nvidia will invest $3 billion in a planned Meta equity offering. Meta plans to use more Nvidia infrastructure for its next-generation AI agents, and the two companies will expand their work on Meta's Muse AI products, covering Vera Rubin systems and Vera CPUs.

Why it matters: Nvidia is putting money into a customer that then buys its chips, an arrangement that fuels debate over whether real demand is being obscured.

Hacker News front page

AI is ruining haunted houses

一位经营鬼屋二十余年的从业者指出,AI 生成图像正渗入鬼屋的宣传海报、场景道具和周边商品,他今年到访的第一家鬼屋就贴出了低质量 AI 海报,页面角落还留着 Gemini 的水印。

Latent Space

Anthropic files for IPO, expected to list in two weeks at roughly $2T valuation

Anthropic has publicly filed its S-1 to list on Nasdaq under the ticker ANT, with the IPO expected in about two weeks and the market expecting a valuation near $2T. The filing shows H1 2026 revenue of $16.3B at an 85% gross margin, and Q2 revenue of $11.6B, up 1,368% year on year, with operating profit of $974M.

Why it matters: Anthropic books cloud marketplace channel fees as sales, marketing and partnership spend, so the cost of the 48% of revenue distributed through cloud partners stays out of its 85% gross margin.

Hacker News front page

The Mythical Man-Month in the Age of AI

文章重新审视《人月神话》中的布鲁克斯定律在 AI 智能体与 LLM 时代是否仍成立:智能体毫秒级上手、不需休息,但顺序任务依然顺序,"九个智能体也无法一个月生出孩子"。开销并未消失,只是从管理人变成管理智能体——仍需有人定义"完成"、审查代码、捕捉模型幻觉并拼接成果;缺乏清晰编排时,多个智能体只会高速循环地互相重写对方的工作。

Hacker News front page

There are now 5 inference providers for abliterated DeepSeek

AllModels 为 abliterated DeepSeek 模型提供统一接入层,目前已有 5 家推理服务商可用,涵盖 DeepSeek V4.1 Flash Abliterated 与 GLM 5.3 等模型。平台通过单一 OpenAI 兼容 API 路由请求,支持自动故障转移、按价格择优以及仅向支持多模态的服务商转发图像请求。

TechCrunch · AI

OpenAI repackages coding tool's agent skills as dots, an always-on personal assistant

OpenAI has repackaged its coding tool's agent capabilities as dots, an always-on personal assistant that keeps working in the background and offers help unprompted. dots launched in late September for Pro and higher ChatGPT desktop and mobile subscribers, with plans starting at $100 a month. It can connect to calendars, email, the computer and other tools.

Why it matters: OpenAI is selling the coding tool's agent skills as a persistent assistant, aiming to cut the prompting and manual steering users do through one continuous conversation and proactive research.