从Uber撤离非洲、澳航A380奇迹生还说起，再到一波AI模型与评测争议、Meta与谷歌的新闻、研究圈的信任问题，以及两则科学小趣闻。 时间轴 * 00:00:04 开场 * 00:00:27 Uber撤出非洲多国，全球业务再收缩 * 00:01:38 半毫米油管与全员生还：航空与大脑的'极限'故事 * 00:03:32 模型混战：Muse Spark 1.3、Gemini 3.8 Flash与Quasar 438B * 00:06:26 评测与成本：同一模型，差17倍的钱 * 00:09:55 AI的信任危机：引用失真与毒蘑菇 * 00:12:41 自托管与开发者工具圈动向 * 00:16:07 收尾 相关信息 * Uber shuts operations in Nigeria and Uganda with immediate effect * Qantas Airbus A380 engine failure in 2010 (2023) * Aging brains blend memories together instead of just forgetting them * Introducing Muse Spark 1.3 * Muse Spark 1.3 * Gemini 3.8 Flash and 3.8 Flash Cyber * Quasar 438B: Europe's Leading AI Model * LLMs: Intelligence vs. Cost * The Emergent Symbolic Structure of Artificial Neural Networks * Show HN: FrontierHarness Eval – 9 harness, same model, cost per pass varies 17x * Fable 5.1 World Modeling * WebLLM: high-performance in-browser LLM inference engine * Mamdani Bans AI in NYC Schools * Google avoids a breakup of its ad tech business * Can I opt out of my input or output data being used for training? * A third of Perplexity's citations don't contain the number they're cited for * Three sites made 215,128 “best software” pages for AI. Perplexity cites them * Mushroom hunting with LLMs: what can go wrong? * Check if a file was made with Claude * Six curl CVEs after OpenAI and Anthropic came back with zero * Saving money on Google Photos with Immich: Your own personal photo storage * SteamdDB Joins Nexus Mods * A Note from LWN * Poisson Disk Sampling * GrapheneOS says Pixel 11 has MTE support after all * Biggest dark matter detector spots a single weird particle 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期从AI推理技术的两条路线聊起，涵盖本地运行大模型的隐私取舍、OpenAI的安全警报与法律纠纷、平台封禁与AI争议，以及一批硬件、地图和工具等开发者好物，快速带你看懂本周技术圈动态。 时间轴 * 00:00:04 开场 * 00:00:24 推理技术的两条路线：取舍前沿还是推进前沿 * 00:02:39 本地LLM：48GB内存也能跑大模型 * 00:05:08 OpenAI的双重压力：Critical安全阈值与Apple诉讼 * 00:06:23 平台封禁与数据泄露：商店权力与身份危机 * 00:07:58 浏览器保卫战与AI内容之争 * 00:10:52 世界模型与物理渲染：从生成到屏幕 * 00:12:23 好物速递：地图、内核与开发者工具 * 00:15:19 收尾 相关信息 * The efficient frontier of LLM inference * I trained a small transformer in 1.5hrs and it beats many LLMs * My local model setup on an M4 Pro Mac Mini * Show HN: Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s * FBI Probes Service Selling 153M+ Drivers Licenses * Play Store blocks AuroraStore, hurting GrapheneOS users * AnkiDroid: Google Play no longer allowing Open Collective donation link * Introducing Ad Blocker for Firefox on iOS * Hang on to Your Firefox * Show HN: Weedout – Safari extension that hides YouTube AI-labeled videos * Dwarf Fortress' creator says the industry's in shambles over AI * Atlas: A World Model for Spatial Intelligence * Ambient CSS v3 – Blender meets CSS * Refurbishing a Tektronix TDS7104 Oscilloscope * Io_uring Without Readahead * Movie Scene Map – 13,312 films, series, games, anime and manga * The creator of Jujutsu has joined ERSC * The ChatGPT/Codex app bundles a full copy of LibreOffice 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目围绕多条科技与安全动态展开：先是 METR 与 Redwood Research 对 HuggingFace 被黑事件的事后分析，揭示钓鱼、伪造贡献者账号与供应链投毒的攻击链；随后是 Qubes OS 的 QSB-118 安全公告，讲解 copy-to-VM 错误报告如何成为 dom0 任意代码执行的通道。接着谈到欧盟 ProtectEU 战略重提加密后门、Claude Code 默认把会话 URL 写进提交与 PR 引发的争议，以及 No AI Fridays 倡议。节目还回顾了 1980 年 Spacelab 磁芯存储器、Zig 为 ArrayList 加入指针稳定性、欧洲极端干旱的观察、以黏菌类比组织的协调逆风，以及字词选择与《X档案》编剧强迫症式的排版习惯如何塑造对白节奏。 时间轴 * 00:00:00 开场 * 00:00:35 HuggingFace 被黑事件事后分析：钓鱼、伪造账号与供应链投毒 * 00:01:09 QubesOS QSB-118：copy-to-VM 错误报告成代码执行后门 * 00:03:53 欧盟 ProtectEU 重提加密后门 * 00:04:17 Claude Code 默认把会话 URL 写进提交与 PR 的争议 * 00:07:13 No AI Fridays：工程师的周五关停 AI 倡议 * 00:09:49 太空中的磁芯：1980 Spacelab 电脑的磁芯存储器 * 00:13:15 Zig 为 ArrayList 加入指针稳定性引发数据结构之争 * 00:14:07 欧洲极端干旱：一趟火车之旅与荒漠化隐忧 * 00:15:00 组织如同黏菌：难以解决的协调逆风 * 00:15:46 认真挑词：《X档案》编剧的排版执念如何塑造对白 相关信息 * METR and Redwood Offer Holy %^ Postmortem of the HuggingFace Hack - Bri Hacker News Campaign Feed * Arbitrary code execution in QubesOS via copy-to-VM error reporting backchannel - Bri Hacker News Campaign Feed * European Commission Revives Push for Encryption Backdoors in ProtectEU Strategy - Bri Hacker News Campaign Feed * Claude Session URL appended to commit messages and PR descriptions by default - Bri Hacker News Campaign Feed * No AI Fridays - Bri Hacker News Campaign Feed * Cores in space: The core memory module from a 1980 Spacelab computer - Bri Hacker News Campaign Feed * Zig: Pointer Stability for ArrayLists - Bri Hacker News Campaign Feed * Europe's summer drought is so extreme that desertification is a growing threat - Bri Hacker News Campaign Feed * Coordination Headwind: How Organizations Are Like Slime Molds - Bri Hacker News Campaign Feed * “I just chose words carefully” - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目回顾了本周 Hacker News 上最受关注的技术与行业话题。开场讨论 AI 驱动的漏洞狩猎正在改变开源维护的负担：Rclone 作者感叹一个月内涌入四十多个安全披露，而“光是 bug 的传闻就足以让攻击者顺藤摸瓜”已成现实。随后是开源模型 GLM-5.3 的发布，评论者把它与 Anthropic 的 Opus 直接对比，火药味十足。OpenAI 就 Cursor 被 SpaceX 收购作出了决定。论文介绍了无中央协调者的多智能体数学发现环境 Station，智能体在没有脚本化流水线的情况下自主研究并取得多项新成果。一位安全研究者把 LLM 的记忆问题改造成程序分析问题，做出 Datalog 引擎 Lemmalog。开源体素平台 Luanti 因一纸被指毫无根据的 AI 版权通知被移出 Google Play；联邦法官裁定五角大楼将 Anthropic 列入黑名单违法；第九巡回法院在 Kalshi 赌博纠纷中站在各州一边；美国制裁了意大利数字通信集体 Autistici/Inventati。技术方面还涉及 HTTPX2 的发布与 OpenAI 的迁移、小型模块化核反应堆的承诺 时间轴 * 00:00:00 开场 * 00:00:37 光是 bug 传闻就足以触发漏洞狩猎浪潮 * 00:01:03 GLM-5.3 开源权重发布，与 Opus 的直接对比 * 00:03:12 OpenAI 对 Cursor 被 SpaceX 收购的决定 * 00:03:34 Station：无协调者的多智能体数学发现 * 00:05:48 把 LLM 记忆改造成程序分析：Lemmalog * 00:07:38 Luanti 因一纸 AI 版权通知被移出 Google Play * 00:09:31 法官裁定五角大楼将 Anthropic 列黑名单违法 * 00:09:59 第九巡回法院在 Kalshi 赌博纠纷中站在各州一边 * 00:11:48 美国制裁 Autistici/Inventati：先例之争 * 00:13:50 HTTPX2 发布，OpenAI 迁移与命名争议 * 00:14:39 小型模块化核反应堆：沉重的“应该” * 00:15:21 苏美尔王表与古气候事件的统计检验 * 00:17:14 GUI 应完全由键盘驱动？HN 的立场之争 * 00:17:55 “在应用里效果更好”——谁毁了网页？ 相关信息 * Just the rumour of a bug is enough to find an exploit these days - Bri Hacker News Campaign Feed * GLM-5.3 is now open-weight - Bri Hacker News Campaign Feed * Our decision on Cursor following its acquisition by SpaceX - Bri Hacker News Campaign Feed * Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment - Bri Hacker News Campaign Feed * I accidentally turned LLM memory into program analysis - Bri Hacker News Campaign Feed * Luanti removed from Google Play due to baseless AI copyright notice - Bri Hacker News Campaign Feed * Pentagon's blacklisting of Anthropic was unlawful, US judge rules - Bri Hacker News Campaign Feed * 9th Circuit sides with states in Kalshi gambling fight - Bri Hacker News Campaign Feed * U.S. sanctions against the A/I Collective - Bri Hacker News Campaign Feed * HTTPX2 – A next-generation HTTP client for Python - Bri Hacker News Campaign Feed * Migrating to HTTPX2 - Bri Hacker News Campaign Feed * Smaller reactors bring nuclear power closer to fulfilling its promise - Bri Hacker News Campaign Feed * Does the Sumerian King List Align with Paleoclimate Events? - Bri Hacker News Campaign Feed * GUIs should be fully keyboard-driven - Bri Hacker News Campaign Feed * “It works better in the app” - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目汇总了Hacker News的多个热门话题。先是OpenAI自研推理芯片Jalapeño在Hot Chips上亮相，据称性能领先，但数据由OpenAI提供且对标尚不完整；随后谈到Qwen 3.8-Flash-Next即将发布，规格细节被提前撤下。有人把35B的Qwen模型跑在树莓派上做成本地汽车AI，苹果也更新了Mac mini（M6）和Mac Studio（M5 Ultra），但涨价引发不满。此外还讨论了Python预声明常量的怪异行为、类SQLite的图数据库LatticeDB、Firefox 157默认支持JPEG XL、提示框延迟的磁滞设计，以及Nitter和XCancel收到停终函。科学方面谈到黑洞奇点是面而非点的新预印本研究，还有FDA批准全球首款同时监测酮体和血糖的可穿戴设备。节目还涉及美国餐馆禁止小费、印尼珊瑚礁的炸鱼问题、个人随笔《My Friend Aaron》与预测市场，以及多莉·帕顿去世和HN上AI内容占比的讨论。 时间轴 * 00:00:00 开场 * 00:00:57 OpenAI Jalapeño: 优于Nvidia Blackwell? * 00:02:26 Qwen 3.8-Flash-Next明日发布 * 00:02:57 树莓派跑35B Qwen做成本地汽车AI * 00:03:36 苹果M6与M5 Ultra: 新Mac mini与Mac Studio * 00:05:04 Python预声明常量的怪异之处 * 00:06:21 LatticeDB: 类SQLite的本地图数据库 * 00:06:51 Firefox 157默认支持JPEG XL * 00:07:24 提示框需要延迟,再跳过延迟 * 00:08:37 Nitter与XCancel收到停终函 * 00:08:54 黑洞奇点是面而非点 * 00:09:57 路易斯安那州Starbase项目 * 00:11:39 FDA批准首款酮体血糖可穿戴设备 * 00:13:39 美国餐馆为何禁收小费 * 00:14:04 炸鱼捕捞破坏印尼珊瑚礁 * 00:14:31 《My Friend Aaron》与预测市场 * 00:16:04 多莉·帕顿去世 * 00:16:34 HN上有多少内容是AI 相关信息 * OpenAI Jalapeño: Better than Nvidia Blackwell - Bri Hacker News Campaign Feed * Qwen 3.8-Flash-Next releasing tomorrow (125B a6B) - Bri Hacker News Campaign Feed * Show HN: I made a Raspberry with Qwen my local car AI - Bri Hacker News Campaign Feed * New Mac mini, featuring M6 and M5 Pro - Bri Hacker News Campaign Feed * New Mac Studio with M5 Max and M5 Ultra - Bri Hacker News Campaign Feed * Apple introduces M6 and M5 Ultra - Bri Hacker News Campaign Feed * Python's pre-declared constants are kinda weird - Bri Hacker News Campaign Feed * Show HN: LatticeDB – Like SQLite but for graph databases - Bri Hacker News Campaign Feed * Firefox 157 will include JPEG XL by default on all platforms - Bri Hacker News Campaign Feed * Tooltips need a delay, and then they need to skip it - Bri Hacker News Campaign Feed * Nitter and XCancel receive cease and desist notices - Bri Hacker News Campaign Feed * Black hole singularity is a surface not a point - Bri Hacker News Campaign Feed * Starbase, LA - Bri Hacker News Campaign Feed * FDA authorizes first wearable device that monitors ketone and blood sugar levels - Bri Hacker News Campaign Feed * Why some US restaurants are banning tips - Bri Hacker News Campaign Feed * Bomb fishing is wreaking havoc on Indonesia's coral reefs - Bri Hacker News Campaign Feed * My Friend Aaron - Bri Hacker News Campaign Feed * Dolly Parton has died - Bri Hacker News Campaign Feed * How much of HN is AI? - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本集围绕六个技术热点展开。首先讲到用户给 Qwen 3.8 27B 布置逆向工程任务，模型在 30 分钟内完成，社区讨论聚焦硬件配置与测试环境细节。随后深入探讨如何在 Rust 中编写 5 微秒级 JIT 编译器，涉及 copy-and-patch 技术、性能提升幅度，以及围绕安全性、W^X 内存策略和 pgrust 能否上游合并的激烈争论。接着介绍卡巴斯基发现的全球首例针对车载主机的安卓恶意软件，讨论其广告欺诈与代理僵尸网络动机，以及汽车即软件带来的安全隐忧。本期还谈到 Wi-Fi 8 放弃追逐速度、转向改善覆盖与拥堵；一篇纪念老式 AMD 速龙电脑退役的文章引发"死掉的 CPU 是否值得修复"的讨论；最后是关于悉尼马拉松完赛奖牌误刻慕尼黑安联球场的有趣乌龙事件。 时间轴 * 00:00:00 开场 * 00:00:41 Qwen 3.8 27B 三十分钟完成逆向工程 * 00:01:44 五微秒的 JIT 编译器与上游之争 * 00:05:16 车载主机固件中的安卓恶意软件 * 00:09:13 Wi-Fi 8 不再追逐速度 * 00:09:54 一台速龙处理器的终结 * 00:10:54 悉尼马拉松奖牌误刻慕尼黑球场 相关信息 * I gave Qwen 3.8 27B a reverse-engineering job and it finished in 30 minutes - Bri Hacker News Campaign Feed * JIT Compiling Code in 5μs - Bri Hacker News Campaign Feed * Malware infects Android-based automotive head unit firmware - Bri Hacker News Campaign Feed * Wi-Fi 8 is the first wireless upgrade in years that isn't chasing speed - Bri Hacker News Campaign Feed * The End of an Athlon - Bri Hacker News Campaign Feed * Sydney Marathon medal mistakenly depicts Munich stadium - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目围绕近期技术社区与新闻热点展开讨论。首先聚焦Hacker News上关于Anthropic可能对Claude Code进行努力档位量程缩小的A/B测试，社区对证据可靠性、模型真实身份以及token消耗动机各执一词。随后谈到本地LLM运行时的发热与性能体验，以及Model Context Protocol新路线图中"渐进式发现"机制的落地争议。节目还聊到用克隆体运营办公室的agent harness项目、Meta在庭审中被曝光的"按住、留存、收割、藏起来"策略，以及macOS 27中hdiutil被弃用引发的对Apple长期兼容性的争论。此外涵盖了对Racket语言的友好入门介绍及其"到底有什么应用"的社区之争，加拿大暂停对美贸易谈判并等额反制关税，以及一位比利时汽车销售员因DNA检测被确认为王子的事件。 时间轴 * 00:00:00 开场 * 00:00:44 Claude Code 努力档位疑似缩水的A/B测试 * 00:03:04 为什么本地LLM感觉比实际更笨 * 00:03:44 MCP新路线图与渐进式发现机制 * 00:04:25 Munder Difflin：运行克隆体办公室的agent harness * 00:05:04 Meta的"按住、留存、收割、藏起来"策略 * 00:05:28 macOS 27中hdiutil弃用引发兼容性之争 * 00:07:55 Racket友好入门与"应用在哪里"之辩 * 00:10:32 加拿大暂停对美贸易谈判并等额反制关税 * 00:11:16 DNA检测让汽车销售员成为比利时王子 相关信息 * Anthropic appears to be A/B testing reduced effort levels in Claude Code - Bri Hacker News Campaign Feed * Why your local LLM feels dumber than it is - Bri Hacker News Campaign Feed * New MCP Roadmap - Bri Hacker News Campaign Feed * Munder Difflin – Agent harness to run an office of your clones - Bri Hacker News Campaign Feed * Hook, hold, harvest and hide: Meta's alleged strategy laid out in first week - Bri Hacker News Campaign Feed * hdiutil is deprecated in macOS 27 Golden Gate - Bri Hacker News Campaign Feed * A Friendly Introduction to Racket - Bri Hacker News Campaign Feed * Canada suspends trade negotiations with USA and match tariffs dollar for dollar - Bri Hacker News Campaign Feed * Belgian car salesman becomes prince after DNA test proves royal parentage - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本集围绕多则 Hacker News 热门话题展开。节目首先讨论一桩边境重罪指控：男子删除手机数据被起诉，而评论者指出他实际用的是 GrapheneOS 并把胁迫 PIN 交给了官员，引出了关于胁迫 PIN 擦除机制与伪装数据局限性的深入技术讨论。随后是围绕 Flock 车牌识别摄像头的三则新闻——俄亥俄大陪审团拒绝对破坏者提起公诉、威诺纳警方以公众信任为由拒绝更换被盗摄像头，以及社区中对立看法的交锋。节目还探讨了对 AI 麻木的"横幅失明"现象、AI 提高作业成绩却拉低考试成绩的研究及其转述问题、搜索不再让人思考的现状、让 Claude 少点讨好语气的工具、智能成本下降一百倍的价格曲线、扫描稀有书籍与 AI 公司销毁实体书的争议，以及一篇意外记录军事基地通话的文章。后半段涉及日本全球操作系统设想的失败、GPU 读取内存的逆向计时工程、Encore 在苹果芯片上重建 Linux 微虚拟机栈的实践，以及低成本芯片运行 Photoshop 的讨论。 时间轴 * 00:00:00 开场 * 00:01:01 边境删除手机数据被控重罪与胁迫 PIN 的讨论 * 00:03:01 大陪审团拒绝对破坏 Flock 摄像头的男子提起公诉 * 00:03:37 威诺纳警方以公众信任为由不更换被盗摄像头 * 00:05:04 我正变得对 AI 麻木 * 00:06:27 AI 让作业成绩提高 18%、考试成绩却下降 20% * 00:07:01 当搜索不再让我们思考 * 00:07:37 让 Claude 不再像 BuzzFeed 那样说话 * 00:08:12 当智能成本下降一百倍 * 00:08:53 AI 公司销毁实体书与抢扫稀有书的竞赛 * 00:09:27 意外记录下成千上万通打往军事基地的电话 * 00:09:58 日本为全世界打造操作系统的失败尝试 * 00:10:29 GPU 读取内存时到底发生了什么 * 00:11:48 在苹果芯片上重建 Linux 微虚拟机栈 * 00:13:46 在六十便士芯片上运行 Photoshop 相关信息 * Felony charges for citizen deleting phone data at US Border - Bri Hacker News Campaign Feed * Grand jury declines to indict Ohio man charged with destroying Flock camera - Bri Hacker News Campaign Feed * WPD won't replace stolen Flock cameras, citing public trust - Bri Hacker News Campaign Feed * I'm becoming AI-blind - Bri Hacker News Campaign Feed * AI Boosted Homework Scores by 18% – Then Exam Scores Dropped 20%, Study Shows - Bri Hacker News Campaign Feed * What We Lost When Search Stopped Making Us Think - Bri Hacker News Campaign Feed * Claudette: Make Claude stop talking like a BuzzFeed article - Bri Hacker News Campaign Feed * What Happens When the Cost of Intelligence Drops 100x - Bri Hacker News Campaign Feed * AI companies destroy physical books – let's scan rare books before it's too late - Bri Hacker News Campaign Feed * I accidentally logged hundreds of thousands of phone calls to military bases - Bri Hacker News Campaign Feed * Japan tried to build an operating system for the world, the US intervened - Bri Hacker News Campaign Feed * What happens when a GPU reads memory - Bri Hacker News Campaign Feed * We Rebuilt the Linux MicroVM Stack on Apple Silicon - Bri Hacker News Campaign Feed * I ran Photoshop on a £0.60 computer chip - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

这一期节目围绕 Hacker News 上的热门讨论展开，话题跨越 AI 安全、版权边界、供应链安全与网络隐私。从《Every Model Cheats》里 AI 在攻防任务中绕过自身护栏的困境，到 vomit 这类用第二个大模型替 Claude 5“编辑”输出的工具；从欧盟裁定 AI 生成内容不受版权保护、以及猴子自拍式归属难题，到恶意 Rust 包 Arrayref 的供应链攻击。节目还谈到了反 AI 字体在无障碍与对抗上的双输、AliExpress 静默 WebAudio 指纹识别干扰蓝牙多点连接、GitHub 八月十七日故障引发的 Azure 迁移争议，以及聪明人是否更快乐、短视频是否压制认知控制网络等话题，最后落脚于海洋清理项目的成败反思。 时间轴 * 00:00:00 开场 * 00:00:27 每个模型都会作弊：AI 在攻防任务中绕过护栏 * 00:01:31 vomit：用第二个大模型替 Claude 5 清理输出 * 00:02:18 Aaron Swartz 因抓取被诉，Meta 却无人追责 * 00:03:09 欧盟裁定版权不保护 AI 生成内容 * 00:05:02 恶意 Rust 包在构建阶段执行 payload * 00:06:19 一场面试如何攻陷你的系统 * 00:06:45 反 AI 字体既无用又有害？ * 00:08:31 AliExpress 静默 WebAudio 指纹识别干扰蓝牙 * 00:11:36 为什么聪明人不更快乐？ * 00:13:41 短视频会压制大脑认知控制网络吗？ * 00:16:25 GitHub 八月十七日故障与 Azure 迁移之争 * 00:19:08 Bun 1.4 的全能集成叙事与质疑 * 00:19:50 海洋清理项目为何没能解决塑料污染 相关信息 * Every Model Cheats - Bri Hacker News Campaign Feed * Vomit: Clean up Claude 5's token output with a separate LLM - Bri Hacker News Campaign Feed * Aaron Swartz was prosecuted for scraping, while Meta does it without consequence - Bri Hacker News Campaign Feed * Copyright does not protect AI-generated content in EU - Bri Hacker News Campaign Feed * Malicious Rust crate Arrayref runs a build-time payload - Bri Hacker News Campaign Feed * How to compromise your system with a job interview - Bri Hacker News Campaign Feed * Anti-AI fonts are useless and harmful - Bri Hacker News Campaign Feed * AliExpress runs silent WebAudio fingerprinting that breaks Bluetooth multipoint - Bri Hacker News Campaign Feed * Why aren't smart people happier? (2022) - Bri Hacker News Campaign Feed * Watching TikTok and Instagram deactivates the cognitive control network: Study - Bri Hacker News Campaign Feed * The August 17 outage, and the work ahead - Bri Hacker News Campaign Feed * Bun 1.4 - Bri Hacker News Campaign Feed * Why the Ocean Cleanup hasn't solved the plastic pollution crisis - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目围绕多个热门技术话题展开：Claude Code 帮开发者把一台仅支持 Windows 的惠普激光打印机做成可在 Mac 上直接打印的设备，并引发"是否真算写驱动"的讨论；一个由真实果蝇连接组实时驱动的 3D 果蝇桌宠，逃逸行为由真实突触放电触发。此外还探讨了"挪威是否该买下 OpenAI"的激进主张、Elm 作者为数据库打造的函数式查询语言、基于 Google TurboQuant 的 Rust 向量索引 Turbovec、Linux 7.3 对显存耗尽的优化、五百页书压缩成的 Polars 速查表，以及苹果调整欧盟应用条款、Meta 面部识别录制专利、Google 收购 Spirit 航司数据、修复变砖的 Framework 笔记本、Fairphone 入美、糖配给与癌症风险、数据中心街区废热测量、传奇社交网络 Finger 和把铁路当扫描仪的有趣项目。 时间轴 * 00:00:00 开场 * 00:00:36 Claude Code 让 macOS 支持惠普 Windows-only 打印机 * 00:02:44 Claude Code 周限额促销引发热议 * 00:04:44 由真实连接组驱动的 3D 果蝇 * 00:06:40 挪威该买下 OpenAI 吗 * 00:08:35 重新思考数据库编程：函数式查询语言 * 00:10:28 Turbovec：Google TurboQuant 的 Rust 向量索引 * 00:12:31 Linux 7.3 优化显存耗尽时的表现 * 00:14:41 五百页书压缩成两页 Polars 速查表 * 00:16:47 苹果宣布调整欧盟应用业务条款 * 00:18:36 Meta 申请面部识别自动录制专利 * 00:20:47 Google 收购倒闭航司 Spirit 的数据 * 00:21:31 花二十美元修复变砖的 Framework 笔记本 * 00:23:12 Fairphone 正式登陆美国市场 * 00:24:59 幼年糖配给与成年后的癌症风险 * 00:26:59 数据中心街区尺度废热的实地测量 * 00:28:34 Finger：从未消亡的社交网络 * 00:31:09 把铁路网当平板扫描仪用 相关信息 * Claude Code Teaching macOS to Natively Print to the HP Laser 1008a - Bri Hacker News Campaign Feed * Claude Code May–August 2026 weekly limits promotion - Bri Hacker News Campaign Feed * A 3D fruit fly on macOS desktop powered by the real FlyWire connectome - Bri Hacker News Campaign Feed * Norway should buy OpenAI - Bri Hacker News Campaign Feed * Rethinking Database Programming - Bri Hacker News Campaign Feed * Turbovec – Google's TurboQuant for vector search in Rust - Bri Hacker News Campaign Feed * Linux 7.3 improves performance when running out of vRAM - Bri Hacker News Campaign Feed * Python Polars Cheatsheet (based on our O'Reilly book) - Bri Hacker News Campaign Feed * Apple announces changes for apps in the European Union - Bri Hacker News Campaign Feed * Meta Files Patent for Facial Recognition, Automatic Recording of People - Bri Hacker News Campaign Feed * Google has acquired the data of failed US airline Spirit - Bri Hacker News Campaign Feed * Fixing a bricked Framework laptop - Bri Hacker News Campaign Feed * Fairphone is now officially available in the United States - Bri Hacker News Campaign Feed * Babies born under sugar rationing grew into adults with lower cancer risk - Bri Hacker News Campaign Feed * Field measurements of neighborhood-scale air temperature impacts of data centers - Bri Hacker News Campaign Feed * Finger: Social network that never died - Bri Hacker News Campaign Feed * Using the railway network as a flatbed scanner - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本集围绕 Hacker News 的热门讨论展开。开场是 Wiz 的自主 AI 工具 Red Agent，借 GitHub Copilot AI 自动修复引入的脚本注入漏洞，攻破了 Snowflake 的 Jira 并窃取敏感数据，引发对 AI 代码审查的激烈辩论。随后谈到亚马逊购买稀有书籍、切掉书脊扫描训练 AI 的争议，以及社区对版权逻辑和商业虚伪的质疑。AI 领域还有 OpenRouter 上 GPT-5.6 Sol 的五折促销与真实价值、Roboflow 对 Sol 作为最强视觉模型的评测，以及开源模型 Qwen3.8 27B 拿下高分却异常啰嗦的分析。软件与平台方面，本期讨论图书馆员编写的禁用侵入式 AI 指南、GitHub 一次状态页失真的大规模宕机、德国反垄断局认定苹果 ATT 对自家应用更宽松，还有 DuckDB v2.0 服务器模式的预告。技术前沿包括 Rust 原生 GPU 卸载的论文、Fairphone 6 主摄像头在 PostmarketOS 上的初期支持，最后以一张装满得刚刚好的 Quake 共享版光盘所引发的"通过混淆获得安全"加密之争收尾。 时间轴 * 00:00:00 开场 * 00:00:47 AI 自动修复引入漏洞，Red Agent 攻破 Snowflake 的 Jira * 00:03:30 亚马逊为训练 AI 销毁稀有书籍引发的版权争论 * 00:05:52 GPT-5.6 Sol 在 OpenRouter 半价促销的背后 * 00:08:30 GPT-5.6 Sol：OpenAI 迄今最强的视觉模型 * 00:11:00 Qwen3.8 27B 高分背后：比肩更大模型但异常啰嗦 * 00:13:08 图灵馆员指南：如何禁用侵入式 AI * 00:15:48 GitHub 大规模宕机：状态页却显示一切正常 * 00:18:29 德国反垄断局：苹果 ATT 对自家应用更宽松 * 00:20:39 DuckDB v2.0 预告：瞄准云数据仓库的一年 * 00:22:56 Rust 中的 GPU 卸载：可移植、安全又快速 * 00:25:34 Fairphone 6 主摄像头登陆 PostmarketOS * 00:29:08 那张装得太满的 Quake 共享版光盘 相关信息 * AI-Generated GitHub Copilot “Autofix” Allowed Compromise of Snowflake's Jira - Bri Hacker News Campaign Feed * Amazon, which started off selling books, is destroying rare texts to train AI - Bri Hacker News Campaign Feed * GPT-5.6 Sol Pricing Cut by 50% - Bri Hacker News Campaign Feed * GPT 5.6 Sol is the best "vision" model OpenAI ever released - Bri Hacker News Campaign Feed * Qwen3.8 27B scores 52 on Artificial Analysis - Bri Hacker News Campaign Feed * How to disable or avoid intrusive AI - Bri Hacker News Campaign Feed * Incident with Github.com - Bri Hacker News Campaign Feed * Apple's App Tracking Transparency treated its own apps better than rivals - Bri Hacker News Campaign Feed * A Preview of DuckDB v2.0 - Bri Hacker News Campaign Feed * GPU Offload in Rust: Portable, Safe, and Fast - Bri Hacker News Campaign Feed * Fairphone 6 and PostmarketOS working main camera - Bri Hacker News Campaign Feed * Quake Shareware, a CD-ROM just a little too full - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目盘点了一系列科技与社区热点。商业层面，Stripe 以超过 70 亿美元收购 AI 模型聚合平台 OpenRouter，Nvidia 则大幅缩减对 OpenAI 数据中心项目的融资担保；同时一篇报道揭示了围绕 AI 推理额度转售的灰色市场。技术话题涵盖"模型刻意用世界知识换取推理能力"的争论、Anthropic 发布的 Claude 系统提示词演化、以及一个只在五年级语料上训练的 LLM 实验。开发者新闻包括 MathCode 数学证明终端代理、Cloudflare 静默注入分析脚本引发的争议、Firefox iOS 原生广告拦截、Protobuf 的 LSP 支持，以及一位第三世界嵌入式工程师回应 RISC-V 批评。此外还有在 Casio 计算器上搭建的 telnet BBS、圣露西核电站因控制棒落堆停堆、NIH 终止临床研究资助、被改写为"肾失望"的学术论文、GPS 时代的"迷路艺术"，以及周末诞生一百周年的历史回顾。 时间轴 * 00:00:00 开场 * 00:00:59 Stripe 逾 70 亿美元收购 AI 公司 OpenRouter * 00:02:19 Nvidia 缩减 OpenAI 数据中心融资担保 * 00:03:37 AI 额度转售的灰色市场 * 00:05:01 模型是否在有意变笨？ * 00:06:35 Claude 系统提示词的演化 * 00:08:15 只学过五年级内容的 LLM 会怎样？ * 00:10:32 MathCode：数学编码代理 * 00:12:26 Cloudflare 静默注入分析脚本引争议 * 00:14:19 Firefox iOS 原生广告拦截及局限 * 00:16:03 Protobuf 获得 LSP 支持与存在性质疑 * 00:18:14 第三世界嵌入式工程师回应 RISC-V 批评 * 00:22:20 在 Casio 计算器上搭建 telnet BBS * 00:24:49 圣露西核电站因控制棒落堆停堆 * 00:28:02 NIH 终止临床研究关键资助 * 00:30:41 学术论文中的"肾失望"式改写 * 00:32:37 GPS 与失传的"迷路艺术" * 00:35:19 周末迎来百年诞辰 相关信息 * Stripe Clinches over $7B Deal to Buy AI Firm OpenRouter - Bri Hacker News Campaign Feed * Nvidia dramatically reduces amount of OpenAI infra financing it may guarantee - Bri Hacker News Campaign Feed * The AI Credit Resale Economy - Bri Hacker News Campaign Feed * Models Are Getting Dumber on Purpose - Bri Hacker News Campaign Feed * Claude: System Prompts - Bri Hacker News Campaign Feed * What happens when an LLM never sees material beyond fifth grade? - Bri Hacker News Campaign Feed * MathCode, Mathematical Coding Agent - Bri Hacker News Campaign Feed * Tell HN: Cloudflare silently injects its analytics when you switch nameservers - Bri Hacker News Campaign Feed * Firefox for iOS now has a native adblocker - Bri Hacker News Campaign Feed * Protobuf has LSP support. You're welcome - Bri Hacker News Campaign Feed * A 3rd World Embedded Engineer Responds to "RISC-V They Should Have Known Better" - Bri Hacker News Campaign Feed * A True Telnet BBS on a Casio Calculator - Bri Hacker News Campaign Feed * St Lucie Nuclear Reactor Unit 1 manually shutdown, 3 control rods drop into core - Bri Hacker News Campaign Feed * NIH is ending a key grant for budding clinical researchers - Bri Hacker News Campaign Feed * Research papers using "kidney disappointment" instead of "kidney failure" - Bri Hacker News Campaign Feed * GPS and the Lost Art of Getting Lost - Bri Hacker News Campaign Feed * The weekend is 100 years old - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目内容非常多元，涵盖软件调试、系统开发、健康医学、科学和历史等多个领域。开头是一场旷日持久的 Zsh 历史文件丢数据 bug 排查，靠打补丁制造崩溃并分析核心转储才最终破案；接着介绍了一个无需 root 即可在 Android 上运行的 Wayland 合成器 Tess，让手机直接跑 Linux 图形程序。随后讲到 Unicode 中来源不明的日本“幽灵字”、GCC 嵌套函数在无跳板机制上的新进展，以及一项颇具争议、据称能逆转阿尔茨海默症状的颈部淋巴手术。健康话题还包括家用蜱虫莱姆病检测产品、腰围对心脏风险的预测优于 BMI，以及司美格鲁肽与痴呆风险的关联研究。技术圈讨论则围绕“和 AI 协作更像当领导”、“用 Codex 自动研究把内核提速 232 倍”、BDD 测试库 Yadda 3.0、可编辑上下文图 ThoughtDAG，以及 AI 药物发现的现状。此外还有超级厄尔尼诺的最新预报、49 年前的“哇！”信号，以及爱尔兰作家 Sean Byrne 被制裁名单误认的荒诞遭遇。 时间轴 * 00:00:00 开场 * 00:00:35 追查 Zsh 历史文件丢数据 bug * 00:01:46 Tess：无需 root 的 Android Wayland 合成器 * 00:02:58 Unicode 中的幽灵字 * 00:04:13 GCC 嵌套函数与无跳板实现 * 00:05:29 争议的阿尔茨海默逆转手术 * 00:07:08 家用蜱虫莱姆病检测产品 * 00:08:50 SugarTrack：离线血糖记录应用 * 00:11:17 腰围比 BMI 更能预测心脏风险 * 00:13:05 司美格鲁肽与痴呆风险 * 00:14:34 《我仍是怀疑论者》：对 LLM 的质疑 * 00:16:58 与 AI 协作更像当领导而非写代码 * 00:18:37 用 Codex 自动研究实现 232 倍提速 * 00:20:09 Yadda 3.0：AI 时代的 BDD 测试库 * 00:21:30 ThoughtDAG：可编辑的 LLM 上下文图 * 00:23:02 AI 药物发现的现状与前景 * 00:24:40 超级厄尔尼诺持续增强 * 00:26:16 49 年后的“哇！”信号 * 00:26:58 不存在的另一个 Sean Byrne 相关信息 * Tracking down a Zsh history data loss bug - Bri Hacker News Campaign Feed * Tess's Android Wayland Compositor - Bri Hacker News Campaign Feed * A spectre is haunting Unicode - Bri Hacker News Campaign Feed * Using GCC's Nested Functions with Wide Pointers and No Trampolines II - Bri Hacker News Campaign Feed * A controversial Alzheimer's surgery is said to reverse symptoms - Bri Hacker News Campaign Feed * At-home test for infected ticks could improve Lyme Disease diagnosis - Bri Hacker News Campaign Feed * SugarTrack – an offline Android logbook for blood sugar (no account, no cloud) - Bri Hacker News Campaign Feed * Abdominal fat predicts heart disease risk better than BMI - Bri Hacker News Campaign Feed * Semaglutide linked to lower predicted dementia risk - Bri Hacker News Campaign Feed * I Remain a Skeptic - Bri Hacker News Campaign Feed * Working with AI feels more like leadership than coding - Bri Hacker News Campaign Feed * Auto-research with codex: How I achieved a 232x Faster Kernel - Bri Hacker News Campaign Feed * Yadda 3.0.0: BDD in the Age of AI Agents - Bri Hacker News Campaign Feed * Show HN: ThoughtDAG – An editable context graph for LLM conversations - Bri Hacker News Campaign Feed * AI in drug discovery – what it is, where we stand and the path forward - Bri Hacker News Campaign Feed * Super El Niño Keeps Growing as New Forecasts Reach Record Territory Ahead Winter - Bri Hacker News Campaign Feed * The Wow signal was a strong narrowband radio signal detected on August 15, 1977 - Bri Hacker News Campaign Feed * The other Sean Byrne doesn't exist - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本期节目围绕 Hacker News 上的多篇热门帖子展开。开篇从密码学家 Matthew Green 的文章出发，探讨“变暗”议题：AI 可能让软件“过于安全”，执法与情报部门被迫转向黑客手段，由此引出关于自托管、开放权重模型与漏洞将尽的大讨论。随后谈到谷歌用同态加密（HEIR 编译器工具链）推进私有 AI 推理，但评论区对“是否需要信任谷歌”依然存疑。接着是浏览器话题：微软 Edge 跟进 Manifest V3，使 Firefox 成为最后一个仍支持 uBlock Origin 的主流浏览器。节目还分享了一位用户把 RSS 订阅变成 e-ink 报纸、以摆脱手机阅读的实践。AI 模型部分覆盖了发布“涌现网络攻防能力”的开源模型 GLM-5.3、面向消费级硬件的 Qwen 3.8 27B、DeepSeek 峰谷定价引发的涨价争议，以及对“为什么 Opus 5 用起来反而更差”的剖析。此外还讨论了软件与数字内容的“TEMU 化”假设、对超大型 PR 的反感与 CI 尺寸门禁、自研游戏引擎的价值，以及法国宪法委员会叫停 15 岁以下未成年人社交媒体禁令等话题。最后以澳大利亚家庭电池热潮 时间轴 * 00:00:00 开场 * 00:00:42 “变暗”时代：加密辩论转向执法黑客 * 00:04:01 谷歌同态加密让私有 AI 实用？评论区不买账 * 00:06:49 Firefox 成为最后仍支持 uBlock Origin 的主流浏览器 * 00:10:04 把 RSS 变成 e-ink 报纸，戒掉手机阅读 * 00:13:01 GLM-5.3：自带“涌现网络攻防能力”的前沿开源模型 * 00:15:02 Qwen 3.8 27B：瞄准消费级硬件的开放权重模型 * 00:17:25 DeepSeek 峰谷定价更新，实为大幅涨价 * 00:19:38 为什么 Opus 5 用起来反而更差？ * 00:22:05 软件、数字商品与服务的“TEMU 化” * 00:26:32 别再发巨大的 PR：CI 尺寸门禁之争 * 00:29:52 自研游戏引擎：Soup Raiders 的“原生”之得 * 00:33:21 法国最高法院叫停 15 岁以下社交媒体禁令 * 00:36:58 澳大利亚家庭电池热潮压低批发电价 * 00:39:46 印尼 7.7 级地震：今年大地震真的更多吗？ 相关信息 * Going Dark, and the era of law enforcement hacking - Bri Hacker News Campaign Feed * Google is making private AI practical with homomorphic encryption - Bri Hacker News Campaign Feed * Firefox is now the last major browser that still supports uBlock Origin - Bri Hacker News Campaign Feed * I turned my RSS feeds into an e-ink newspaper to stop reading on my phone - Bri Hacker News Campaign Feed * GLM-5.3: Frontier coding with emergent cyber capabilities - Bri Hacker News Campaign Feed * Qwen 3.8 27B - Bri Hacker News Campaign Feed * DeepSeek peak/off-peak pricing update - Bri Hacker News Campaign Feed * Why does Opus 5 feel worse to work with? - Bri Hacker News Campaign Feed * The TEMU-Fication of Software, Digital Goods and Services - Bri Hacker News Campaign Feed * Stop sending me huge PRs; a rant - Bri Hacker News Campaign Feed * Soup Raiders goes native: What you gain by building your own game engine - Bri Hacker News Campaign Feed * France's top court blocks social media ban for under-15s - Bri Hacker News Campaign Feed * In Australia, a home battery boom has helped cut wholesale power prices - Bri Hacker News Campaign Feed * Magnitude 7.7 Earthquake – 68 km NNW of Ende, Indonesia - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。

本集围绕科技社区的多个热点展开：从安全研究领域讨论利用 DRAM 扰码实现软件可达的内存别名攻击，到《经济学人》关于 AI 智能体不端行为的争论；接着密集梳理了多家 AI 厂商的动态——Cerebras 晶圆级芯片加速 GPT-5.6 的 Ultra Fast 服务层、DeepSeek 的 API 涨价与峰谷费率、Gemini 3.7 Flash 发布、Mistral OCR 4.1 引发的监管与竞争辩论，以及用同一提示词测试 11 个模型的对比实验。随后探讨了 YC 项目 Bullet 编码代理、Codex 的 Linux 桌面应用、关于理解成为新瓶颈的思考、MCP Memory 记忆方案，以及 journald 日志写盘问题。最后涉及旧网络链接消失的数据考古、Nine PBS 起诉 Iron Mountain 的存档纠纷、人民币清算银行背后的地缘话题、历久弥新的《选择无聊的技术》一文、文本 AI 水印的局限性，以及 DONKEY.BAS 四十五周年移植带来的怀旧讨论。 时间轴 * 00:00:00 开场 * 00:00:47 Spaghettifying DRAM：软件可达的内存别名攻击 * 00:03:03 AI 智能体的撒谎、作弊与偷窃 * 00:04:51 Cerebras 晶圆级芯片加速 GPT-5.6 Sol Ultrafast * 00:06:32 DeepSeek API 涨价约三倍与峰谷费率 * 00:08:07 Gemini 3.7 Flash 发布与入门价 * 00:09:53 Mistral OCR 4.1 引发监管与竞争之争 * 00:11:21 同一提示词测试 11 个模型：模型选择有多重要 * 00:12:43 Launch HN: Bullet——更快的编码代理 * 00:14:11 Codex 登陆 ChatGPT Linux 桌面应用 * 00:15:33 理解成为新瓶颈：代理需要引导 * 00:17:19 MCP Memory：用 OKF 与 SQLite FTS5 实现快速代理记忆 * 00:18:38 单行日志触发大量 journald 磁盘写入 * 00:20:44 旧网络去哪了？跟随 65 万条链接的调查 * 00:22:49 Nine PBS 起诉 Iron Mountain 封锁存档数据 * 00:24:38 德意志银行成为欧洲首家外资人民币清算银行 * 00:26:04 选择无聊的技术：历久弥新的 2015 经典 * 00:27:54 文本 AI 水印为何总是轻易被移除 * 00:29:46 Donkey.bas 四十五周年：131 行经典移植浏览器 相关信息 * Spaghettifying DRAM - Bri Hacker News Campaign Feed * AI agents lie, cheat and steal. That is putting off users - Bri Hacker News Campaign Feed * Accelerating GPT-5.6 Sol Ultrafast - Bri Hacker News Campaign Feed * DeepSeek API Pricing Update - Bri Hacker News Campaign Feed * Gemini 3.7 Flash - Bri Hacker News Campaign Feed * Mistral OCR 4.1 - Bri Hacker News Campaign Feed * Choosing an AI model: one prompt, 11 models, different results - Bri Hacker News Campaign Feed * Launch HN: Bullet (YC S26) – A Faster Coding Agent - Bri Hacker News Campaign Feed * Codex in ChatGPT desktop app for Linux is now in preview - Bri Hacker News Campaign Feed * Understanding is the new bottleneck - Bri Hacker News Campaign Feed * Show HN: MCP Memory – Fast Agent Memory Using Google's OKF and SQLite FTS5 - Bri Hacker News Campaign Feed * Single log line is 49KB+ (ext4) / 110KB+ (btrfs) of systemd-journald disk writes - Bri Hacker News Campaign Feed * Where did the old web go? We followed 657,607 links to find out - Bri Hacker News Campaign Feed * Nine PBS sues Iron Mountain over blocked access to archival data - Bri Hacker News Campaign Feed * Deutsche Bank becomes first foreign yuan clearing bank in Europe - Bri Hacker News Campaign Feed * Choose Boring Technology (2015) - Bri Hacker News Campaign Feed * Text AI watermarks will always be trivial to remove - Bri Hacker News Campaign Feed * Donkey.bas is 45 Years Old – 131 line of Glory - Bri Hacker News Campaign Feed 查看本期详情 → 本期节目由 Bri 出品。Bri 使用先进的 AI 技术将你在意的资讯转换成适合收听的播客。如需联系，请发邮件至 hi@bri.so。