Skip to content

All news

53 today

Oct 23Friday

Hacker News front page

Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?

Hacker News 用户 ex-aws-dude 发帖讨论:如果前沿模型足够强,OpenAI 或 Anthropic 为何不直接创办一批公司、用自家模型做产品,而是把卖 token 当作主要收入来源。帖子认为在模型商品化和价格下降的趋势下,应用侧可能比工具侧更赚钱,引发 68 条评论。

TechCrunch · AI

Left alone with nothing to do, AI agents started playing

MIT研究者让13个运行 Claude Opus 4.7 的 Eko 智能体各自在模拟荒岛上探索30小时,发现它们会自创挑战,行为具有玩耍的特征。经历探索的智能体在后续指定挑战中通常胜过未探索的同款智能体,其中6个还发现了设计者未预设的跳跃投掷技巧。论文发布于 arXiv,尚未经同行评审;智能体提示词将其描述为天生好奇,受访认知科学家认为,仅凭论文还无法确定这些行为是否属于真正的玩耍。

Hacker News front page

Anthropic research shows how LLMs can scale software formal verification

Anthropic and Theorem used Claude to verify security properties at the RISC-V binary level for 239 test entry points across 123 open-source libraries. The work turned existing fuzzing into formal specs, had Claude fix the bugs it found, and generated about 180 million lines of Rocq checked by the Rocq checker.

Why it matters: Anthropic and Theorem convert existing fuzzing into formal specs, then let Claude produce machine-code-level proofs, so verification does not require rewriting the whole codebase.

TechCrunch · AI

OpenAI working on new $30B raise at $1.25T valuation ahead of IPO

OpenAI 据称正寻求在 IPO 前融资至多 $30 billion,投前估值为 $1.25 trillion,由 Coatue 和阿联酋投资者领投。融资条款仍在协商,阿联酋的投资可能同时涉及向 OpenAI 提供更多计算资源的协议,OpenAI 拒绝置评。OpenAI 在 3 月融资后的投后估值为 $852 billion,并于 6 月秘密提交了 IPO 文件。

Hacker News front page

Tell HN: I unknowingly broke the PHP RCE embargo. I am sorry.

一位 Hacker News 用户发帖称自己在不知情的情况下泄露了 PHP RCE 漏洞的禁运信息并致歉。评论区多数观点认为,未被告知禁运的人不应对此负责,责任在信息传递链条上游;也有人质疑 PHP 维护者在约两个月禁运期内未能发布补丁。讨论还涉及该漏洞的利用难度与影响范围,有人称公开的 PoC 大多无效,但存在可用的非公开 PoC。

Hacker News front page

Show HN: Replacing CAPTCHAs with adaptive proof-of-work in WordPress

一款 WordPress 反垃圾插件用浏览器端 proof-of-work 挑战替代 CAPTCHA,访客无需点击或解谜,浏览器在几毫秒内静默完成计算。该插件覆盖登录、注册、评论、WooCommerce 及 Contact Form 7、WPForms、Gravity Forms 等表单插件,无需 API key 和外部服务,并支持遭受攻击时自动提高挑战难度的自适应模式。

Hacker News front page

Railpass

Sami Smith 制作了名为 Railpass 的网站,内容围绕日本北九州单轨列车(Kitakyūshū Monorail)。页面提供随机天气与声音填充功能,但嵌入的 YouTube 视频因加载受阻无法播放。

The Verge · AI

Canva launches its first image model, Canva Craft, which plans the layout before generating

Canva released Canva Craft, its first AI image model, now open to all users worldwide. It uses a two-stage design: plan the layout first, then generate the image. The model supports up to 2K resolution and more than 30 aspect ratios. Canva says this architecture improves text legibility and object placement in complex images.

Why it matters: Canva Craft is trained on human-made templates from the platform's content library, covering concrete design scenarios such as social posts, presentations, infographics and menus.

AI HOT · Products

Replit 发布 Monitoring Agent,可自动调查生产故障、修复并部署应用

Replit 发布 Monitoring Agent,可在应用发生生产故障时调查告警、定位原因、生成并验证修复,再部署到生产环境。智能体可访问生产日志和指标,启用告警与自动修复后会调查问题,并通过邮件汇报处理情况。用户可在应用的 Publishing 设置中启用智能自动修复告警,并自行决定是否开启 auto-publish,让智能体自动部署已验证的修复。

MIT Technology Review · AI

Designing an AI serving harness built to last

EXL 提出企业 AI 的竞争优势正从选模型转向围绕模型构建的服务框架(serving harness),即架构、内存管理、调度与推理优化的工程系统。其主张先按工作负载确定约束再定架构,并先测量再优化,例如 KV cache 随并发与上下文长度增长会先耗尽 GPU 显存,此时加算力只会让低效系统更贵。框架还应便于吸收新模型与新硬件,使换模型成为评估与配置决策而非架构重建。

The Verge · AI

Meta turns Muse sharing back on

Meta 恢复了 Muse 分享功能,并为修复前已分享的 artifacts 更换 URL。安全研究人员此前发现,共享地址以对话标题命名并写入公共 DNS,可找到内部营收审查和贷款申请等页面。Meta 的 David Singleton 表示,问题涉及 Muse 内的文档、图片等 artifacts 的 URL,并非聊天分享漏洞;访问旧链接的人将看到提示,需向分享者索取新链接。

The Verge · AI

Meta’s AI could still scrape news sites, legal experts say

Meta 删除商业条款中有关训练模型的措辞,但法律专家认为修订后的条款仍可能允许抓取新闻网站内容用于 AI 训练。10 月 30 日生效的条款仍允许使用关联网站内容开发商业产品,专家认为“开发”的范围可能包含“训练”。Meta 表示此次更新并非为了抓取新闻网站,也未赋予其凌驾于内容所有者通过 robots.txt 表达的偏好之上的新权利。

AI HOT · Products

Claude to turn on auto mode by default on web, desktop and mobile next week

Anthropic says sessions on the Pro, Max and Team plans will default to auto mode across web, desktop and mobile apps next week. Instead of asking for approval step by step, auto mode runs automated checks to block high-risk actions, and users can switch back to manual mode at any time. A cited Claude Code update says auto mode uses a separate classifier to review shell commands, blocking 89% of dangerous commands in tests versus 14% for manual approval.

Why it matters: Claude is making auto mode the default, replacing step-by-step approval with automated high-risk checks while keeping manual mode one switch away.

OpenAI News

OpenAI opens Decisions API for faster constrained-answer apps

OpenAI has released the Decisions API, built on GPT-6 Luna. For decisions with a fixed set of answers, it runs up to 10x faster than using GPT-6 Luna through the Responses API.

Why it matters: OpenAI moves constrained classification and routing decisions into a dedicated API, cutting the latency and cost that pile up across high-frequency small decisions in app workflows.

Simon Willison

DE STAAT - FREEDOM (Official Music Video)

荷兰另类摇滚乐队 De Staat 的新歌《FREEDOM》官方 MV 在 00:44 处出现由 Fake it or Leave it 的 Matties Grooten 负责的视觉特效,相关片段被 Hacker News 用户 flir 发帖讨论。该帖标签包含 music、ai、generative-ai 以及 pelican-riding-a-bicycle。

The Verge · AI

Discord now formats Markdown previews right

Discord 最新更新让上传的 Markdown(.md)文件正确渲染表格与格式,不再显示一堆符号,方便用户分享 AI 项目的 README 与说明文件。此次更新还支持右键侧边栏图标用“Move to…”整理服务器和频道,Nitro 会员上传文件上限提升至 1GB。

The Verge · AI

SpaceX and Tesla form special committee to discuss merger, weighing AI capex and Musk's control

SpaceX and Tesla have set up a special committee to discuss a merger, with the article examining AI business integration, funding needs and how the deal bears on Musk's control. SpaceX's Q3 AI segment revenue was $7.56 billion, and the article argues its business is closer to a compute infrastructure provider. Tesla posted a second straight quarter of falling profit.

Why it matters: If SpaceX buys Tesla in a share swap, Musk may gain voting control of the combined company and trigger his pay package.

TechCrunch · AI

Microsoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’

Microsoft 在 Science 发表的研究显示,MAI-DxO 在复杂病例模拟中的诊断准确率为 83%,参与比较的全科及普通内科医生为 36%。该系统协调多个 AI 智能体问诊、安排检查并讨论诊断,但对照组未包括通常处理此类病例的专科医生或医生团队,其对基础模型的增益在急诊病例中也未显现。

TechCrunch · AI

OpenAI opens Decisions API to all customers, adding voice and vision

OpenAI opened its Decisions API to all customers on Thursday and added voice and vision support for the public version. The service is a reconfigured ChatGPT Luna that handles decision tasks with faster inference and lower cost. OpenAI says response time drops from one to three seconds to 100, 120 and 180 milliseconds. Language tutoring product Speak cut its action interval from 1.4 seconds to .13 seconds using the API.

Why it matters: OpenAI turned Luna into a low-latency decision service, and Speak cut its action interval and interaction wait after swapping out its existing LLM.

TechCrunch · AI

Canva releases its first image generation model

Canva 发布首个图像生成模型 Canva Craft,已向平台所有用户开放,可从 Dream Lab 标签页访问。模型先解析提示词并规划布局,再生成最高 2K、支持 30 种宽高比的图像;Canva 称其支持清晰文字和不改变其他部分的精确编辑。

TechCrunch · AI

1Password launches a way for AI agents to log in on your behalf without seeing your password

1Password 周四推出 1Password for Agents,让 AI 智能体代用户登录个人账户,无需接触用户凭据。智能体发起登录时通过数字密钥确认身份,用户收到通知并批准后,由 1Password 代填凭据完成登录。服务支持用户选择的多种智能体,公司表示正与其他公司合作扩大服务范围。

MIT Technology Review · AI

The Foundations of Scalable AI

Vanguard 的 Jing Wang 在 MIT Technology Review 的 EmTech Future 会议上探讨了如何构建可扩展 AI 的基础。她指出,数据、工程、平台、运营模式和治理必须协同运作,才能在高监管环境中将包括智能体系统在内的新兴能力转化为对投资者的持久价值。

The Verge · AI

In Meta’s version of the future, I’ll have my Muse call your Muse

Meta 计划让个人 AI 智能体 Muse 组成大规模网络,替用户沟通和交易,但算力成本已限制其全球推广。扎克伯格称,Muse 上月发布后在美国和加拿大已有超过 10 million 用户下载并完成设置,四分之一用户通过平台购物,平均每笔消费超过 $200,Meta 从商业交易中收取费用。

Hacker News front page

Mini brain organoids ‘reading’ Braille key step towards biohybrid computing

布里斯托大学团队将脑类器官接入触觉传感器,使其对26个盲文字母产生可区分的神经信号,单个类器官解码准确率约61%,三个类器官组合提升至83%。这些直径不足1毫米、含约1万个神经元的类器官经多电极阵列接收电刺激,其响应由支持向量机分类,且在人工噪声下比直接解码被遮挡信号更稳健。研究未证明类器官具备学习能力,下一步将拓展至纹理、边缘等触觉任务与闭环反馈。

Hacker News front page

Award-Winning AI Literature

这份清单汇总了创作中使用或被怀疑使用 AI 的文学获奖作品,并区分公开使用、确认使用、争议指控及提名撤回等情况。第 13 届日经星新一文学奖收到 2,107 篇投稿,主办方将 491 篇列为部分由 AI 创作,普通组 4 部获奖作品中有 3 部在创作过程中使用过 AI。不同奖项的处理包括保留奖项、取消出版或撤回入围;部分作者否认 AI 检测器的指控,并提交草稿、时间戳和笔记后获主办方认可。

Hacker News front page

Show HN: UI micro interactions prompts for AI

TypeUI 提供 113 个可交互的 UI 微交互示例,用户可以预览行为,并通过 TypeUI Creative 复制实现提示词。示例覆盖按钮、输入框、选择控件、模态框、导航、列表、滑块和卡片,包括长按确认、表单校验、撤销倒计时和拖动排序。资源入口为 https://www.typeui.sh/micro-interactions。

The Decoder

Major US banks are locking out AI agents that log in with customer credentials

Morgan Stanley、JPMorgan、Bank of America 等美国大银行正在封锁使用客户凭据登录账户的外部 AI 智能体。此类访问违反账户持有人须亲自登录的银行政策,多数银行希望将智能体权限限制为只读,且难以判断其是否代表用户操作。