1
谷歌推出 AI Ultra 新订阅计划，月费 100 美元起
谷歌推出月费100美元新订阅计划，瞄准中高端AI用户，填补Pro与Ultra之间的市场空白。
IT之家 5 月 20 日消息，在今日的 2026 谷歌 I/O 开发者大会上，谷歌 CEO 桑达尔 · 皮查伊（Sundar Pichai）宣布推出 100 美元的新订阅计划，支持 Gemini Spark AI 智能体。 该订阅计划瞄准 AI Pro（月费 20 美元）与 AI Ultra（月费…
2
AMD AI 开发者大会首次在中国举行，苏姿丰出席，共谈 AI 智能体新趋势
AMD首次在中国举办AI开发者大会，苏姿丰与李开复共话AI智能体新趋势，端侧计算与多智能体成焦点。
IT之家 5 月 19 日消息，今天，AMD AI 开发者大会（AMD AI DevDay 2026）在上海正式举办，这是该面向全球 AI 开发者的技术大会首次登陆中国，IT之家也受邀前往现场，首先一大感受就是人气火爆，参会的人数远比想象中要多。 具体来看，这次大会以“开放 - 构建 - 创新 - …
3
Google Makes It Easy to Deepfake Yourself
Google推出新“avatar”功能，让你无需亲自出镜就能用AI生成自己的数字分身，创作更轻松。
Google’s overhaul of its AI creation software, Flow, includes a new video model and a tool for generating selfie videos called avatars.
4
Run Claude Managed Agents with Vercel Sandbox
Vercel Sandbox 让 Claude 托管代理在隔离微 VM 中运行，企业级安全与性能
You can now run Claude Managed Agents with Vercel Sandbox . Claude Managed Agents handles the model, harness, tools, and session state. Self-hosting l…
5
In stunning display of stupid, secret CISA credentials found in public GitHub repo
CISA大量明文密码、SSH密钥等敏感凭证在GitHub公共仓库中暴露数月，安全失误令人震惊。
SSH keys, plaintext passwords, other sensitive data had been up since November 2025.
6
Google updates its Gemini app to take on ChatGPT and Claude at IO 2026
Google IO 2026上Gemini重大更新，Daily Brief与AI视频模型直击ChatGPT和Claude。
The updates signal Google’s push to turn its Gemini app into an all-purpose AI hub rather than a stand-alone chatbot.
7
Design ain’t a democracy
民主决策反成产品开发枷锁，揭示科技公司效率低下的根源在于不敢做决定。
Democracies are the worst way to build a product. This is the reason why most tech companies are slow and ineffective: managers aren’t really there to…
8
Against Landlords
房东擅自闯入让人愤怒，这篇文章犀利剖析租房者的无奈与不公。
The other day our landlord walked into our apartment. Just opened the door right up and headed in. They were here to fix a problem we’d never heard ab…
9
Keyboard-first apps
键盘优先应用正流行：这些工具通过硬核模式强制禁鼠标，帮你彻底掌握快捷键！
I don’t know if this is a trend or anything but I like the idea of this: keyboard-first apps. Chris shared Godspeed , a todo list app, that has a “har…
10
Blockchain is not only crappy technology but a bad vision for the future
知名评论家犀利指出：区块链不仅技术糟糕，其构建的未来充满不信任与偏执，是加密中世纪地狱
This piece by Kai Stinchcombe on why blockchain is a terrible idea is pretty dang quotable—he tackles the crazy idea that blockchain is a magical tech…
11
What would happen if we just gave people money?
深入探讨全民基本收入（UBI）的研究与乌托邦理想，揭示工作意义的核心困境。
This is a sharp and brilliant piece by Andrew Flowers. He looks at the current research behind universal basic income, where the government would effe…
12
Quickness and Detours
卡尔维诺论“快”与迂回：文学之美不在直线，而在变幻的藏身处
In Six Memos for the Next Millennium Italo Calvino outlines all of the attributes and properties of great writing that he believed ought to thrive int…
13
Like Father, Like Daughter: Continuing Jerry Gomez’s Legacy
女儿深情追忆父亲招牌绘画生涯，传承技艺与热爱的动人故事。
Paige Gomez on piecing together her late father’s legacy — and following in his footsteps.
14
Heuristic Pathologies and Further Variance Reduction via Uncertainty Propagation in the AIVAT Family of Techniques
论文揭示AIVAT方差缩减技术中的启发式病态，并提出不确定性传播进一步降低方差，提升多智能体性能评估精度。
arXiv:2605.14261v1 Announce Type: new Abstract: How should an agent's performance in a multiagent environment be evaluated when there is a limited sam…
15
TAPIOCA: Why Task- Aware Pruning Improves OOD model Capability
探究任务感知剪枝如何提升模型在分布外数据上的表现，揭示内在机制
arXiv:2605.14738v1 Announce Type: cross Abstract: Recent work has promoted task-aware layer pruning as a way to improve model performance on particula…
16
FactorizedHMR: A Hybrid Framework for Video Human Mesh Recovery
FactorizedHMR提出混合框架，巧妙区分躯干与四肢的不确定性，显著提升视频人体网格恢复在遮挡场景下的精度。
arXiv:2605.14854v1 Announce Type: cross Abstract: Human Mesh Recovery (HMR) is fundamentally ambiguous: under occlusion or weak depth cues, multiple 3…
17
Positive Alignment: Artificial Intelligence for Human Flourishing
跳出传统AI安全框架，探讨如何让人工智能积极促进人类繁荣，开启对齐研究新范式。
arXiv:2605.10310v2 Announce Type: replace Abstract: Existing alignment research is dominated by concerns about safety and preventing harm: safeguards,…
18
Numerical exploration of the range of shape functionals using neural networks
用神经网络参数化凸体，探索形状泛函不等式范围，数学优化新框架。
arXiv:2602.14881v2 Announce Type: replace-cross Abstract: We introduce a novel numerical framework for the exploration of Blaschke--Santal\'o diagrams…
19
Running OpenClaw on Amazon EC2 with Claude and Telegram
在EC2上搭建基于Claude的Telegram机器人，手把手教你用OpenClaw实现智能交互。
Article URL: https://blog.harun.dev/running-openclaw-on-amazon-ec2-with-claude-and-telegram Comments URL: https://news.ycombinator.com/item?id=4799804…
20
Loop Earplugs Discount Codes: 40% Off
Loop耳塞低至六折，舒适降噪不损音质，音乐节常备神器，折扣码别错过！
Save on Loop Earplugs, including Quiet 2 and popular gift sets for improved sleep, focus, and comfort.