// Sat, 09 May 2026 03:53:27 +0000 Sat, 09 May 2026 03:53:27 +0000 Jekyll v3.10.0 <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>Simon Willison：</strong>I recently talked with Joseph Ruscio about AI coding tools for Heavybit's High Leverage podcast: Ep. #9, The AI Coding Paradigm Sh…</p> <p><strong>Simon Willison：</strong>Behind the Scenes Hardening Firefox with Claude Mythos Preview Fascinating, in-depth details on how Mozilla used their access to t…</p> <p><strong>Simon Willison：</strong>There weren't a lot of big new announcements from Anthropic at yesterday's Code w/ Claude event, but the biggest by far was the de…</p> <p><strong>Simon Willison：</strong>Using Claude Code: The Unreasonable Effectiveness of HTML Thought-provoking piece by Thariq Shihipar (on the Claude Code team at A…</p> <p><strong>Simon Willison：</strong>Release: llm-gemini 0.31 gemini-3.1-flash-lite is no longer a preview Here's my write-up of the Gemini 3.1 Flash-Lite Preview mode…</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Tool: GitHub Repo Stats One of the things I alwa…｜中文观点：GitHub Repo Stats 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性…</p> <p><strong>总结 + 观点：</strong>Tool: Big Words I'm using my vibe coded macOS pr…｜中文观点：Big Words 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</p> <p><strong>总结 + 观点：</strong>WebRTC is designed to degrade and drop my prompt…｜中文观点：围绕 Quoting Luke Curley，真正重要的是它会不会影响团队的模型选型、性能…</p> <p><strong>总结 + 观点：</strong>Release: datasette-referrer-policy 0.1 The OpenS…｜中文观点：datasette-referrer-policy 0.1 的核心不在新鲜感，而在它是否能…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Vibe coding and agentic engineering are getting closer than I'd like</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：I recently talked with Joseph Ruscio about AI coding tools for Heavybit's High Leverage podcast: Ep. #9, The AI Coding Paradigm Shift with Simon Willison Here are some of my highlights, including my disturbing realization that vibe coding and agentic engineering have started to converge in my own work. One thing I really enjoy about podcasts is that they sometimes push me to think out loud in a way that exposes an idea I've not previously been able to put into words. Vibe coding and agentic engineering are starting to overlap A few weeks after vibe coding was first coined I published Not all AI-assisted programming is vibe coding (but vibe coding rocks) where I firmly staked out my belief that "vibe coding" is a very different beast from responsible use of AI to write code, which I've since started to call agentic engineering When Joseph brought up the distinction between the two I had a sudden realization that they're not nearly as distinct for me as they used to be: Weirdly though, those things have started to blur for me already, which is quite upsetting. I thought we had a very clear delineation where vibe coding is the thing where you're not looking at the code at all. You might not even know how to program. You might be a non-programmer who asks for a thing, and gets a thing, and if the thing works, then great! And if it doesn't, you tell it that it doesn't work and cross your fingers. But at no point are you really caring about the code quality or any of those additional constraints. And my take on vibe coding was that it's fantastic, provided you understand when it can be used and when it can't. A personal tool for you, where if there's a bug it hurts only you, go ahead! If you're building software for other people, vibe coding is grossly irresponsible because it's other people's information. Other people get hurt by your stupid bugs. You need to have a higher level than that. This contrasts with agentic engineering where you are a professional software engineer. You understand security and maintainability and operations and performance and so forth. You're using these tools to the highest of your own ability. I'm finding the scope of challenges I can take on has gone up by a significant amount because I've got the support of these tools. But I'm still leaning on my 25 years of experience as a software engineer. The goal is to build high quality production systems: if you're building lower quality stuff faster, I think that's bad. I want to build higher quality stuff faster. I want everything I'm building to be better in every way than it was before. The problem is that as the coding agents get more reliable, I'm not reviewing every line of code that they write anymore, even for my production level stuff. I know full well that if you ask Claude Code to build a JSON API endpoint that runs a SQL query and outputs the results as JSON, it's just going to do it right. It's not going to mess that up. You have it add automated tests, you have it add documentation, you know it's going to be good. But I'm not reviewing that code. And now I've got that feeling of guilt: if I haven't reviewed the code, is it really responsible for me to use this in production? The thing that really helps me is thinking back to when I've worked at larger organizations where I've been an engineering manager. Other teams are building software that my team depends on. If another team hands over something and says, "hey, this is the image resize service, here's how to use it to resize your images"... I'm not going to go and read every line of code that they wrote. I'm going to look at their documentation and I'm going to use it to resize some images. And then I'm going to start shipping my own features. And if I start running into problems where the image resizer thing appears to have bugs or the performance isn't good, that's when I might dig into their Git repositories and see what's going on. But for the most part I treat that as a semi-black box that I don't look at until I need to. I'm starting to treat the agents in the same way. And it still feels uncomfortable, because human beings are accountable for what they do. A team can build a reputation. I can say "I trust that team over there. They built good software in the past. They're not going to build something rubbish because that affects their professional reputations." Claude Code does not have a professional reputation! It can't take accountability for what it's done. But it's been proving itself anyway - time and time again it's churning out straightforward things and doing them right in the style that I like. There's an element of the normalization of deviance here - every time a model turns out to have written the right code without me monitoring it closely there's a risk that I'll trust it at the wrong moment in the future and get burned. The new challenge of evaluating software It used to be if you found a GitHub repository with a hundred commits and a good readme and automated tests and stuff, you could be pretty sure that the person writing that had put a lot of care and attention into that project. And now I can knock out a git repository with a hundred commits and a beautiful readme and comprehensive tests of every line of code in half an hour! It looks identical to those projects that have had a great deal of care and attention. Maybe it is as good as them. I don't know. I can't tell from looking at it. Even for my own projects, I can't tell. So I realized what I value more than the quality of the tests and documentation is that I want somebody to have used the thing. If you've got a vibe coded thing which you have used every day for the past two weeks, that's much more valuable to me than something that you've just spat out and hardly even exercised. The bottlenecks have shifted If you can go from producing 200 lines of code a day to 2,000 lines of code a day, what else breaks? The entire software development lifecycle was, it turns out, designed around the idea that it takes a day to produce a few hundred lines of code. And now it doesn't. It's not just the downstream stuff, it's the upstream stuff as well. I saw a great talk by Jenny Wen who's the design leader at Anthropic, where she said we have all of these design processes that are based around the idea that you need to get the design right - because if you hand it off to the engineers and they spend three months building the wrong thing, that's catastrophic. There's this whole very extensive design process that you put in place because that design results in expensive work. But if it doesn't take three months to build, maybe the design process can be a whole lot riskier because cost, if you get something wrong, has been reduced so much. Why I'm still not afraid for my career When I look at my conversations with the agents, it's very clear to me that this is moon language for the vast majority of human beings. There are a whole bunch of reasons I'm not scared that my career as a software engineer is over now that computers can write their own code, partly because these things are amplifiers of existing experience. If you know what you're doing, you can run so much faster with them. I'm constantly reminded as I work with these tools how hard the thing that we do is. Producing software is a ferociously difficult thing to do. And you could give me all of the AI tools in the world and what we're trying to achieve here is still really difficult. Matthew Yglesias, who's a political commentator, yesterday tweeted "Five months in, I think I've decided that I don't want to vibecode I want professionally managed software companies to use AI coding assistance to make more/better/cheaper software products that they sell to me for money." And that feels about right to me. I can plumb my house if I watch enough YouTube videos on plumbing. I would rather hire a plumber. On the threat to SaaS providers of companies rolling their own solutions instead: I just realized it's the thing I said earlier about how I only want to use your side project if you've used it for a few weeks. The enterprise version of that is I don't want a CRM unless at least two other giant enterprises have successfully used that CRM for six months. You want solutions that are proven to work before you take a risk on them. Tags: ai generative-ai llms podcast-appearances vibe-coding coding-agents agentic-engineering</p> <p>链接：<a href="https://simonwillison.net/2026/May/6/vibe-coding-and-agentic-engineering/#atom-everything" target="_blank">https://simonwillison.net/2026/May/6/vibe-coding-and-agentic-engineering/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Vibe coding and agentic engineering are getting closer than... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Behind the Scenes Hardening Firefox with Claude Mythos Preview</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：Behind the Scenes Hardening Firefox with Claude Mythos Preview Fascinating, in-depth details on how Mozilla used their access to the Claude Mythos preview to locate and then fix hundreds of vulnerabilities in Firefox: Suddenly, the bugs are very good Just a few months ago, AI-generated security bug reports to open source projects were mostly known for being unwanted slop. Dealing with reports that look plausibly correct but are wrong imposes an asymmetric cost on project maintainers: it’s cheap and easy to prompt an LLM to find a “problem” in code, but slow and expensive to respond to it. It is difficult to overstate how much this dynamic changed for us over a few short months. This was due to a combination of two main factors. First, the models got a lot more capable. Second, we dramatically improved our techniques for harnessing these models steering them, scaling them, and stacking them to generate large amounts of signal and filter out the noise. They include some detailed bug descriptions too, including a 20-year old XSLT bug and a 15-year-old bug in the <legend> element. A lot of the attempts made by the harness were blocked by Firefox's existing defense-in-depth measures, which is reassuring. Mozilla were fixing around 20-30 security bugs in Firefox per month through 2025. That jumped to 423 in April. Via Lobste.rs Tags: firefox mozilla security ai generative-ai llms anthropic claude ai-security-research&lt;/p&gt; <p>链接：<a href="https://simonwillison.net/2026/May/7/firefox-claude-mythos/#atom-everything" target="_blank">https://simonwillison.net/2026/May/7/firefox-claude-mythos/#atom-everything</a></p> &lt;/div&gt; <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Behind the Scenes Hardening Firefox with Claude Mythos Previ... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> &lt;/div&gt; &lt;/div&gt; <div class="brief-section"> <div class="brief-title"> <h3>Notes on the xAI/Anthropic data center deal</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #engineering-value</p> <p>作者： </p> <p>原文：There weren't a lot of big new announcements from Anthropic at yesterday's Code w/ Claude event, but the biggest by far was the deal they've struck with SpaceX/xAI to use "all of the capacity of their Colossus data center". As I mentioned in my live blog of the keynote that's the one with the particularly bad environmental record The gas turbines installed to power the facility initially ran without Clean Air Act permits or pollution control devices, which they got away with by classifying them as "temporary". Credible reports link it to increases in hospital admissions relating to low air quality. Andy Masley, one of the most prolific voices pushing back against misleading rhetoric about data centers (see The AI water issue is fake and Data center land issues are fake had this to say about Colossus: I would simply not run my computing out of this specific data center I get that Anthropic are severely compute-constrained, but in a world where the very existence of "AI data centers" is a red-hot political issue (see recent news out of Utah for a fresh example), signing up with this particular data center is a really bad look. There was a lot of initial chatter about how this meant xAI were clearly giving up on their own Grok models, since all of their capacity would be sold to Anthropic instead. That was a misconception - Anthropic are getting Colossus 1, but xAI are keeping their larger Colossus 2 data center for their own work. As an interesting side note, the night before the Anthropic announcement, xAI sent out a deprecation notice for Grok 4.1 Fast and several other models providing just two weeks' notice before shutdown, reported here by @xlr8harder from SpeechMap: This is terrible @xai. I just spent time and money to migrate to grok 4.1 fast, and you're disabling it with less than two weeks notice, after releasing it in November, with no migration path to a fast/cheap alternative. I will never depend on one of your products again. Here's SpeechMap's detailed explanation of how they selected Grok 4.1 Fast for their project in March. Were xAI serving those models out of Colossus 1? xAI owner Elon Musk (who previously delighted in calling Anthropic "Misanthropic" tweeted the following: By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2. And then shortly afterwards Just as SpaceX launches hundreds of satellites for competitors with fair terms and pricing, we will provide compute to AI companies that are taking the right steps to ensure it is good for humanity. We reserve the right to reclaim the compute if their AI engages in actions that harm humanity. Presumably the criteria for "harm humanity" are decided by Elon himself. Sounds like a new form of supply chain risk for Anthropic to me! Tags: ai llms anthropic ai-ethics ai-energy-usage xai andy-masley</p> <p>链接：<a href="https://simonwillison.net/2026/May/7/xai-anthropic/#atom-everything" target="_blank">https://simonwillison.net/2026/May/7/xai-anthropic/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Notes on the xAI/Anthropic data center deal 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Using Claude Code: The Unreasonable Effectiveness of HTML</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：Using Claude Code: The Unreasonable Effectiveness of HTML Thought-provoking piece by Thariq Shihipar (on the Claude Code team at Anthropic) advocating for HTML over Markdown as an output format to request from Claude. The article is crammed with interesting examples (collected on this site and prompt suggestions like this one: Help me review this PR by creating an HTML artifact that describes it. I'm not very familiar with the streaming/backpressure logic so focus on that. Render the actual diff with inline margin annotations, color-code findings by severity and whatever else might be needed to convey the concept well. I've been defaulting to asking for most things in Markdown since the GPT-4 days, when the 8,192 token limit meant that Markdown's token-efficiency over HTML was extremely worthwhile. Thariq's piece here has caused me to reconsider that, especially for output. Asking Claude for an explanation in HTML means it can drop in SVG diagrams, interactive widgets, in-page navigation and all sorts of other neat ways of making the information more pleasant to navigate. I wrote about Useful patterns for building HTML tools last December, but that was focused very much on interactive utilities like the ones on my tools.simonwillison.net site. I'm excited to start experimenting more with rich HTML explanations in response to ad-hoc prompts. Trying this out on copy.fail copy.fail describes a recently discovered Linux security exploit, including a proof of concept distributed as obfuscated Python. I tried having GPT-5.5 create an HTML explanation of the exploit like this: curl https://copy.fail/exp llm -m gpt-5.5 -s 'Explain this code in detail. Reformat it, expand out any confusing bits and go deep into what it does and how it works. Output HTML, neatly styled and using capabilities of HTML and CSS and JavaScript to make the explanation rich and interactive and as clear as possible' Here's the resulting HTML page It's pretty good, though I should have emphasized explaining the exploit over the Python harness around it. Tags: html security markdown ai prompt-engineering generative-ai llms llm claude-code</p> <p>链接：<a href="https://simonwillison.net/2026/May/8/unreasonable-effectiveness-of-html/#atom-everything" target="_blank">https://simonwillison.net/2026/May/8/unreasonable-effectiveness-of-html/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Using Claude Code: The Unreasonable Effectiveness of HTML 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>llm-gemini 0.31</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #ecosystem-shift</p> <p>作者： </p> <p>原文：Release: llm-gemini 0.31 gemini-3.1-flash-lite is no longer a preview Here's my write-up of the Gemini 3.1 Flash-Lite Preview model back in March. I don't believe this new non-preview model has changed since then. Tags: llm-release gemini llm google generative-ai ai llms</p> <p>链接：<a href="https://simonwillison.net/2026/May/7/llm-gemini/#atom-everything" target="_blank">https://simonwillison.net/2026/May/7/llm-gemini/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 llm-gemini 0.31 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>GitHub Repo Stats</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：Tool: GitHub Repo Stats One of the things I always look for when evaluating a new GitHub repository is the number of commits it has... but that number isn't visible on GitHub's mobile site layout. I built this tool to fix that, using this prompt: Given a GitHub repo URL or foo/bar repo ID show information about that repo absorbed via wither REST or graphql CORS fetch() including the number of commits in the repo and other useful stats Example output for simonw/datasette and simonw/llm Tags: github</p> <p>链接：<a href="https://simonwillison.net/2026/May/7/github-repo-stats/#atom-everything" target="_blank">https://simonwillison.net/2026/May/7/github-repo-stats/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">GitHub Repo Stats 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Big Words</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #learning-value</p> <p>作者： </p> <p>原文：Tool: Big Words I'm using my vibe coded macOS presentations tool to put together a talk, and I wanted to add a slide with some text on it. The tool only accepts URLs, so I put together a quick page that accepts query string arguments and turns them into a simple slide. Here's an example: https://tools.simonwillison.net/big-words?text=simonwillison.net&amp;gradient=1&amp;size=9.5 Double click or double tap the page to access a form for modifying the different options. Tags: vibe-coding tools</p> <p>链接：<a href="https://simonwillison.net/2026/May/7/big-words/#atom-everything" target="_blank">https://simonwillison.net/2026/May/7/big-words/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Big Words 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Quoting Luke Curley</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #engineering-value</p> <p>作者： </p> <p>原文：WebRTC is designed to degrade and drop my prompt during poor network conditions. wtf my dude WebRTC aggressively drops audio packets to keep latency low. If you’ve ever heard distorted audio on a conference call, that’s WebRTC baybee. The idea is that conference calls depend on rapid back-and-forth, so pausing to wait for audio is unacceptable. …but as a user, I would much rather wait an extra 200ms for my slow/expensive prompt to be accurate. After all, I’m paying good money to boil the ocean, and a garbage prompt means a garbage response. It’s not like LLMs are particularly responsive anyway. But I’m not allowed to wait It’s impossible to even retransmit a WebRTC audio packet within a browser; we tried at Discord. The implementation is hard-coded for real-time latency or else -- Luke Curley OpenAI’s WebRTC Problem, in response to How OpenAI delivers low-latency voice AI at scale Tags: webrtc openai</p> <p>链接：<a href="https://simonwillison.net/2026/May/9/luke-curley/#atom-everything" target="_blank">https://simonwillison.net/2026/May/9/luke-curley/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Quoting Luke Curley，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>datasette-referrer-policy 0.1</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #ecosystem-shift</p> <p>作者： </p> <p>原文：Release: datasette-referrer-policy 0.1 The OpenStreetMap tiles on the Datasette global-power-plants demo weren't displaying correctly. This turned out to be caused by two bugs. The first is that the CAPTCHA I added to that site a few weeks ago was triggering for the .json fetch requests used by the map plugin, and since those weren't HTML the user was not being asked to solve them. Here's the fix The second was that OpenStreetMap quite reasonably block tile requests from sites that use a Referrer-Policy: no-referrer header. Datasette does this by default, and I didn't want to change that default on people without warning - so I had Codex GPT-5.5 build me a new plugin to help set that header to another value. Tags: openstreetmap http datasette</p> <p>链接：<a href="https://simonwillison.net/2026/May/5/datasette-referrer-policy/#atom-everything" target="_blank">https://simonwillison.net/2026/May/5/datasette-referrer-policy/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">datasette-referrer-policy 0.1 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> </legend></p></div></div></div> Sat, 09 May 2026 00:00:00 +0000 //2026/05/09/AI-News-Weekly-2026-W19.html //2026/05/09/AI-News-Weekly-2026-W19.html AI-Weekly <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>How OpenAI runs Codex securely with sandboxing, approvals, network policies, and agent-native telemetry to support safe and compli…</p> <p><strong>OpenAI Blog：</strong>OpenAI expands Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber, helping verified defenders accelerate vulnerability resear…</p> <p><strong>OpenAI Blog：</strong>Parloa leverages OpenAI models to power scalable, voice-driven AI customer service agents, enabling enterprises to design, simulat…</p> <p><strong>OpenAI Blog：</strong>Explore new realtime voice models in the OpenAI API that can reason, translate, and transcribe speech, enabling more natural and i…</p> <p><strong>OpenAI Blog：</strong>OpenAI begins testing ads in ChatGPT to support free access, with clear labeling, answer independence, strong privacy protections,…</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Introducing Trusted Contact in ChatGPT, an optio…｜中文观点：围绕 Introducing Trusted Contact in ChatGPT，真正重…</p> <p><strong>总结 + 观点：</strong>Simplex boosts software development with ChatGPT…｜中文观点：对 Simplex rethinks software development with…</p> <p><strong>总结 + 观点：</strong>Learn how ChatGPT safeguards your privacy, reduc…｜中文观点：围绕 How ChatGPT learns about the world while p…</p> <p><strong>总结 + 观点：</strong>Uber uses OpenAI to power AI assistants and voic…｜中文观点：围绕 Uber uses OpenAI to help people earn smart…</p> <p><strong>总结 + 观点：</strong>OpenAI’s B2B Signals research shows how frontier…｜中文观点：对 How frontier firms are pulling ahead，更该看它能不…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Running Codex safely at OpenAI</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How OpenAI runs Codex securely with sandboxing, approvals, network policies, and agent-native telemetry to support safe and compliant coding agent adoption.</p> <p>链接：<a href="https://openai.com/index/running-codex-safely" target="_blank">https://openai.com/index/running-codex-safely</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Running Codex safely at OpenAI 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI expands Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber, helping verified defenders accelerate vulnerability research and protect critical infrastructure.</p> <p>链接：<a href="https://openai.com/index/gpt-5-5-with-trusted-access-for-cyber" target="_blank">https://openai.com/index/gpt-5-5-with-trusted-access-for-cyber</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cy... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Parloa builds service agents customers want to talk to</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Parloa leverages OpenAI models to power scalable, voice-driven AI customer service agents, enabling enterprises to design, simulate, and deploy reliable, real-time interactions.</p> <p>链接：<a href="https://openai.com/index/parloa" target="_blank">https://openai.com/index/parloa</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Parloa builds service agents customers want to talk to，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Advancing voice intelligence with new models in the API</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Explore new realtime voice models in the OpenAI API that can reason, translate, and transcribe speech, enabling more natural and intelligent voice experiences.</p> <p>链接：<a href="https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api" target="_blank">https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Advancing voice intelligence with new models in the API，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Testing ads in ChatGPT</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI begins testing ads in ChatGPT to support free access, with clear labeling, answer independence, strong privacy protections, and user control.</p> <p>链接：<a href="https://openai.com/index/testing-ads-in-chatgpt" target="_blank">https://openai.com/index/testing-ads-in-chatgpt</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Testing ads in ChatGPT，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing Trusted Contact in ChatGPT</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Introducing Trusted Contact in ChatGPT, an optional safety feature that notifies someone you trust if serious self-harm concerns are detected.</p> <p>链接：<a href="https://openai.com/index/introducing-trusted-contact-in-chatgpt" target="_blank">https://openai.com/index/introducing-trusted-contact-in-chatgpt</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Introducing Trusted Contact in ChatGPT，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Simplex rethinks software development with Codex</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Simplex boosts software development with ChatGPT Enterprise and Codex, reducing design, build, and testing time while scaling AI-driven workflows.</p> <p>链接：<a href="https://openai.com/index/simplex" target="_blank">https://openai.com/index/simplex</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Simplex rethinks software development with Codex，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How ChatGPT learns about the world while protecting privacy</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how ChatGPT safeguards your privacy, reduces personal data in training, and gives you control over whether your conversations improve AI models.</p> <p>链接：<a href="https://openai.com/index/how-chatgpt-protects-privacy" target="_blank">https://openai.com/index/how-chatgpt-protects-privacy</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 How ChatGPT learns about the world while protecting privacy，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Uber uses OpenAI to help people earn smarter and book faster</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Uber uses OpenAI to power AI assistants and voice features that help drivers earn smarter and riders book faster across a global real-time marketplace.</p> <p>链接：<a href="https://openai.com/index/uber" target="_blank">https://openai.com/index/uber</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Uber uses OpenAI to help people earn smarter and book faster，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How frontier firms are pulling ahead</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI’s B2B Signals research shows how frontier enterprises deepen AI adoption, scale Codex-powered agentic workflows, and build durable competitive advantage.</p> <p>链接：<a href="https://openai.com/index/introducing-b2b-signals" target="_blank">https://openai.com/index/introducing-b2b-signals</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 How frontier firms are pulling ahead，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing Advanced Account Security</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Introducing Advanced Account Security: phishing-resistant login, stronger recovery, and enhanced protections to safeguard sensitive data and prevent account takeover.</p> <p>链接：<a href="https://openai.com/index/advanced-account-security" target="_blank">https://openai.com/index/advanced-account-security</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Introducing Advanced Account Security 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Where the goblins came from</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How goblin outputs spread in AI models: timeline, root cause, and fixes behind personality-driven quirks in GPT-5 behavior.</p> <p>链接：<a href="https://openai.com/index/where-the-goblins-came-from" target="_blank">https://openai.com/index/where-the-goblins-came-from</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Where the goblins came from 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Building the compute infrastructure for the Intelligence Age</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI scales Stargate to build the compute infrastructure powering AGI, adding new data center capacity to meet growing AI demand.</p> <p>链接：<a href="https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age" target="_blank">https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Building the compute infrastructure for the Intelligence Age 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/realtime-voice-component</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：openai/realtime-voice-component recently updated repository.</p> <p>链接：<a href="https://github.com/openai/realtime-voice-component" target="_blank">https://github.com/openai/realtime-voice-component</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/realtime-voice-component 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/jukebox</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Code for the paper "Jukebox: A Generative Model for Music"</p> <p>链接：<a href="https://github.com/openai/jukebox" target="_blank">https://github.com/openai/jukebox</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，openai/jukebox 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>GNUtrition 0.33.0rc1 Now Available</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://lists.gnu.org/archive/html/info-gnutrition/2026-05/msg00000.html Comments URL: https://news.ycombinator.com/item?id=48071610 Points: 1 Comments: 0</p> <p>链接：<a href="https://lists.gnu.org/archive/html/info-gnutrition/2026-05/msg00000.html" target="_blank">https://lists.gnu.org/archive/html/info-gnutrition/2026-05/msg00000.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">GNUtrition 0.33.0rc1 Now Available 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The React2Shell Story and What Happened Next.js</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://sylvie.fyi/posts/react2shell/ Comments URL: https://news.ycombinator.com/item?id=48071582 Points: 1 Comments: 1</p> <p>链接：<a href="https://sylvie.fyi/posts/react2shell/" target="_blank">https://sylvie.fyi/posts/react2shell/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The React2Shell Story and What Happened Next.js 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>ExitFlare</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://exitflare.com/ Comments URL: https://news.ycombinator.com/item?id=48071551 Points: 3 Comments: 0</p> <p>链接：<a href="https://exitflare.com/" target="_blank">https://exitflare.com/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">ExitFlare 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Learn Something Old Every Day, Part XXI: VGA Memory Access Is Complicated</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.os2museum.com/wp/learn-something-old-every-day-part-xxi-vga-memory-access-is-complicated/ Comments URL: https://news.ycombinator.com/item?id=48071548 Points: 1 Comments: 0</p> <p>链接：<a href="https://www.os2museum.com/wp/learn-something-old-every-day-part-xxi-vga-memory-access-is-complicated/" target="_blank">https://www.os2museum.com/wp/learn-something-old-every-day-part-xxi-vga-memory-access-is-complicated/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Learn Something Old Every Day, Part XXI: VGA Memory Access I... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Ovlt Lightweight Auth Server That Runs in ~20MB RAM</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.ovlt.tech/ Comments URL: https://news.ycombinator.com/item?id=48071546 Points: 1 Comments: 1</p> <p>链接：<a href="https://www.ovlt.tech/" target="_blank">https://www.ovlt.tech/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Ovlt – Lightweight Auth Server That Runs in ~20MB RAM 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Sat, 09 May 2026 00:00:00 +0000 //2026/05/09/AI-News-Daily-Issue-70.html //2026/05/09/AI-News-Daily-Issue-70.html AI-Daily Research Model <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>OpenAI updates the Agents SDK with native sandbox execution and a model-native harness, helping developers build secure, long-runn…</p> <p><strong>OpenAI Blog：</strong>OpenAI expands its Trusted Access for Cyber program, introducing GPT-5.4-Cyber to vetted defenders and strengthening safeguards as…</p> <p><strong>OpenAI Blog：</strong>Cloudflare brings OpenAI’s GPT-5.4 and Codex to Agent Cloud, enabling enterprises to build, deploy, and scale AI agents for real-w…</p> <p><strong>OpenAI Blog：</strong>Learn how sales teams use ChatGPT to research accounts, personalize outreach, manage deals, and improve pipeline and conversion.</p> <p><strong>OpenAI Blog：</strong>Learn how to use ChatGPT, start your first conversation, and discover simple ways to write, brainstorm, and solve problems with AI…</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Someone recently suggested to me that the reason…｜中文观点：围绕 R to @karpathy: Someone recently suggested…</p> <p><strong>总结 + 观点：</strong>Judging by my tl there is a growing gap in under…｜中文观点：从 Judging by my tl there is a growing gap in…</p> <p><strong>总结 + 观点：</strong>Karpathy 新发布的最小 ChatGPT 复现项目，训练到推理的完整栈只有几千行可读代码，…｜中文观点：nanochat 最值得看的不是性能，而是它第一次把 ChatGPT 训练+推理的全流程压…</p> <p><strong>总结 + 观点：</strong>Karpathy 早期的教学级 GPT 实现，代码短到可以一口气读完，长期用作理解 Transf…｜中文观点：minGPT 的价值不是生产就绪，而是教材级清晰：它最适合那些想从零搭一遍训练循环、确认自…</p> <p><strong>总结 + 观点：</strong>Anthropic 公开其内部工程师 take-home 面试题，可作为理解他们工程品味和评估标…｜中文观点：这条的信号不是题目本身，而是 Anthropic 把招聘标准开放出来，对想了解他们工程文化…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The next evolution of the Agents SDK</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI updates the Agents SDK with native sandbox execution and a model-native harness, helping developers build secure, long-running agents across files and tools.</p> <p>链接：<a href="https://openai.com/index/the-next-evolution-of-the-agents-sdk" target="_blank">https://openai.com/index/the-next-evolution-of-the-agents-sdk</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The next evolution of the Agents SDK 的价值在于它是否能真正降低智能体落地门槛，而不是再提供一层概念包装。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Trusted access for the next era of cyber defense</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI expands its Trusted Access for Cyber program, introducing GPT-5.4-Cyber to vetted defenders and strengthening safeguards as AI cybersecurity capabilities advance.</p> <p>链接：<a href="https://openai.com/index/scaling-trusted-access-for-cyber-defense" target="_blank">https://openai.com/index/scaling-trusted-access-for-cyber-defense</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Trusted access for the next era of cyber defense 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Enterprises power agentic workflows in Cloudflare Agent Cloud with OpenAI</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Cloudflare brings OpenAI’s GPT-5.4 and Codex to Agent Cloud, enabling enterprises to build, deploy, and scale AI agents for real-world tasks with speed and security.</p> <p>链接：<a href="https://openai.com/index/cloudflare-openai-agent-cloud" target="_blank">https://openai.com/index/cloudflare-openai-agent-cloud</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Enterprises power agentic workflows in Cloudflare Agent Clou... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>ChatGPT for sales teams</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how sales teams use ChatGPT to research accounts, personalize outreach, manage deals, and improve pipeline and conversion.</p> <p>链接：<a href="https://openai.com/academy/sales" target="_blank">https://openai.com/academy/sales</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 ChatGPT for sales teams，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Getting started with ChatGPT</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to use ChatGPT, start your first conversation, and discover simple ways to write, brainstorm, and solve problems with AI.</p> <p>链接：<a href="https://openai.com/academy/getting-started" target="_blank">https://openai.com/academy/getting-started</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Getting started with ChatGPT，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-technical pe...</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：X Andrej Karpathy</p> <p>标签：#x_profiles #extended</p> <p>作者： </p> <p>原文：Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-technical people (who otherwise only knew AI as synonymous with ChatGPT as a website) experienced the latest agentic models.</p> <p>链接：<a href="https://twitter.com/karpathy/status/2042341482531864741" target="_blank">https://twitter.com/karpathy/status/2042341482531864741</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 R to @karpathy: Someone recently suggested to me that the re...，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use.</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：X Andrej Karpathy</p> <p>标签：#x_profiles #extended</p> <p>作者： </p> <p>原文：Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code. But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much value. The goldmines are elsewhere, and the focus comes along. So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions. TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are. staysaasy (@staysaasy) The degree to which you are awed by AI is perfectly correlated with how much you use AI to code. https://nitter.net/staysaasy/status/2042063369432183238#m</p> <p>链接：<a href="https://twitter.com/karpathy/status/2042334451611693415" target="_blank">https://twitter.com/karpathy/status/2042334451611693415</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Judging by my tl there is a growing gap in understanding of... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/nanochat</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Karpathy 新发布的最小 ChatGPT 复现项目，训练到推理的完整栈只有几千行可读代码，目标是把“百美元跑一个 ChatGPT”压到个人可动手的范围。</p> <p>链接：<a href="https://github.com/karpathy/nanochat" target="_blank">https://github.com/karpathy/nanochat</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">nanochat 最值得看的不是性能，而是它第一次把 ChatGPT 训练+推理的全流程压到个人能读懂、能跑通的粒度，对想吃透底层的开发者最有价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/minGPT</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Karpathy 早期的教学级 GPT 实现，代码短到可以一口气读完，长期用作理解 Transformer 训练与推理最短路径的入口。</p> <p>链接：<a href="https://github.com/karpathy/minGPT" target="_blank">https://github.com/karpathy/minGPT</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">minGPT 的价值不是生产就绪，而是教材级清晰：它最适合那些想从零搭一遍训练循环、确认自己真的理解 GPT 的工程师。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/original_performance_takehome</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Anthropic 公开其内部工程师 take-home 面试题，可作为理解他们工程品味和评估标准的一手材料。</p> <p>链接：<a href="https://github.com/anthropics/original_performance_takehome" target="_blank">https://github.com/anthropics/original_performance_takehome</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">这条的信号不是题目本身，而是 Anthropic 把招聘标准开放出来，对想了解他们工程文化与评价尺度的人非常有用。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/symphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI 新开源的多 agent 编排框架，重点不是写代码的 coding agent，而是任务隔离、委派与团队级协作。</p> <p>链接：<a href="https://github.com/openai/symphony" target="_blank">https://github.com/openai/symphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Symphony 的定位更像工作流基础设施：真正的价值在于它把“多 agent 协作”的实现细节标准化，而不是又出一个 coding agent。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-cookbook</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI 官方示例库更新，通常折射出他们希望开发者优先采用的新模式（tool use、structured output、responses API 等）。</p> <p>链接：<a href="https://github.com/openai/openai-cookbook" target="_blank">https://github.com/openai/openai-cookbook</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">cookbook 的更新值得单独跟踪：它折射出 OpenAI 想让开发者默认走哪些新 API 和用法路径，是路线图的早期信号。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/rustbpe</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Karpathy 用 Rust 重写的 BPE tokenizer 训练器，把 tiktoken 里不透明的训练流程变成可学习、可实验的代码。</p> <p>链接：<a href="https://github.com/karpathy/rustbpe" target="_blank">https://github.com/karpathy/rustbpe</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">rustbpe 补上了 tokenizer 训练这块的“黑盒”：它让 tokenizer 变体实验、教学与复现都更直观，研究者最先受益。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-python</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI 官方 Python SDK 更新，通常先于公告暴露出新接口细节、参数变化或默认路径调整。</p> <p>链接：<a href="https://github.com/openai/openai-python" target="_blank">https://github.com/openai/openai-python</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">官方 SDK 的 commit 经常是 API 方向的早期指示灯，对做集成和多模型平台的团队比市场通稿更有参考价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-demos</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Anthropic 官方的 Claude Agent SDK 示例仓库，覆盖代码 agent、文件编辑、工具链编排等典型用法。</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-demos" target="_blank">https://github.com/anthropics/claude-agent-sdk-demos</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">demos 仓库往往比文档更早暴露 SDK 的边界和推荐模式，对正在选型 agent 栈的团队是最值得先跑一遍的材料。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Sightings</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：/elsewhere/sightings/ I have a new camera (a Canon R6 Mark II) so I'm taking a lot more photos of birds. I share my best wildlife photos on iNaturalist and based on yesterday's successful prototype I decided to add those to my blog. I built this feature on my phone using Claude Code for web, as an extension of my beats system for syndicating external content. Here's the PR and prompt. As with my other forms of incoming syndicated content sightings show up on the homepage, the date archive pages, and in site search results. I back-populated over a decade of iNaturalist sightings, which means you that if you search for lemur you'll see my lemur photos from Madagascar in 2019! Tags: blogging photography wildlife ai inaturalist generative-ai llms ai-assisted-programming claude-code</p> <p>链接：<a href="https://simonwillison.net/2026/May/2/sightings/#atom-everything" target="_blank">https://simonwillison.net/2026/May/2/sightings/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Sightings 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>AI Engineer World's Fair Autoresearch, Memory, World Models, Tokenmaxxing, Agentic Commerce, and Vertical AI Call for Speakers</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Latent Space</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：a quiet day lets us make a call for speakers!</p> <p>链接：<a href="https://www.latent.space/p/ainews-ai-engineer-worlds-fair-autoresearch" target="_blank">https://www.latent.space/p/ainews-ai-engineer-worlds-fair-autoresearch</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">[AINews] AI Engineer World's Fair — Autoresearch, Memory, Wo... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Ask HN: Any nice project ideas that you know you'll never bring to life</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Every now and then a nice web/mobile application pop into my head. But because of my day job and a side project I’m currently working on, I already know that this idea will never come to life. Do you just write them down and move on and maybe eventually forget about it? Share them publicly? If you share them and someone else brings the idea to life, would that make you happy? Curious what's your approach? Comments URL: https://news.ycombinator.com/item?id=47981506 Points: 1 Comments: 0</p> <p>链接：<a href="https://news.ycombinator.com/item?id=47981506" target="_blank">https://news.ycombinator.com/item?id=47981506</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Ask HN: Any nice project ideas that you know you'll never br... 本身是信号而不是结果，真正要看的是它把哪些工程和产品判断推向前沿。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>New study finds task switching raises risk in transplant surgeries</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://news.vt.edu/articles/2026/04/pamplin-bit-research-organ-transplant-task-switching.html Comments URL: https://news.ycombinator.com/item?id=47981486 Points: 1 Comments: 0</p> <p>链接：<a href="https://news.vt.edu/articles/2026/04/pamplin-bit-research-organ-transplant-task-switching.html" target="_blank">https://news.vt.edu/articles/2026/04/pamplin-bit-research-organ-transplant-task-switching.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">New study finds task switching raises risk in transplant sur... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>GameStop is preparing offer for eBay</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://finance.yahoo.com/markets/stocks/articles/gamestop-preparing-offer-ebay-wsj-212703455.html Comments URL: https://news.ycombinator.com/item?id=47981465 Points: 1 Comments: 0</p> <p>链接：<a href="https://finance.yahoo.com/markets/stocks/articles/gamestop-preparing-offer-ebay-wsj-212703455.html" target="_blank">https://finance.yahoo.com/markets/stocks/articles/gamestop-preparing-offer-ebay-wsj-212703455.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">GameStop is preparing offer for eBay 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Fri, 08 May 2026 00:00:00 +0000 //2026/05/08/AI-News-Daily-Issue-69.html //2026/05/08/AI-News-Daily-Issue-69.html AI-Daily Agent Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>Learn how Codex helps you go beyond chat by automating tasks, connecting tools, and producing real outputs like docs and dashboard…</p> <p><strong>OpenAI Blog：</strong>Learn how to get started with Codex by setting up projects, creating threads, and completing your first tasks with step-by-step gu…</p> <p><strong>OpenAI Blog：</strong>Explore the GPT-5.5 Bio Bug Bounty: a red-teaming challenge to find universal jailbreaks for bio safety risks, with rewards up to…</p> <p><strong>X Andrej Karpathy：</strong>Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you…</p> <p><strong>OpenAI Blog：</strong>OpenAI makes ChatGPT for Clinicians free for verified U.S. physicians, nurse practitioners, and pharmacists, supporting clinical c…</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Workspace agents in ChatGPT are Codex-powered ag…｜中文观点：对 Introducing workspace agents in ChatGPT，更该看…</p> <p><strong>总结 + 观点：</strong>Learn how to build, use, and scale workspace age…｜中文观点：对 Workspace agents，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而…</p> <p><strong>总结 + 观点：</strong>A deep dive into the Codex agent loop, showing h…｜中文观点：对 Speeding up agentic workflows with WebSocke…</p> <p><strong>总结 + 观点：</strong>OpenAI Privacy Filter is an open-weight model fo…｜中文观点：Introducing OpenAI Privacy Filter 更值得从实际采用价值来…</p> <p><strong>总结 + 观点：</strong>ChatGPT Images 2.0 introduces a state-of-the-art…｜中文观点：比起表面参数，Introducing ChatGPT Images 2.0 更需要观察它是…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>What is Codex?</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how Codex helps you go beyond chat by automating tasks, connecting tools, and producing real outputs like docs and dashboards.</p> <p>链接：<a href="https://openai.com/academy/what-is-codex" target="_blank">https://openai.com/academy/what-is-codex</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">What is Codex? 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How to get started with Codex</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to get started with Codex by setting up projects, creating threads, and completing your first tasks with step-by-step guidance.</p> <p>链接：<a href="https://openai.com/academy/codex-how-to-start" target="_blank">https://openai.com/academy/codex-how-to-start</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">How to get started with Codex 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>GPT-5.5 Bio Bug Bounty</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Explore the GPT-5.5 Bio Bug Bounty: a red-teaming challenge to find universal jailbreaks for bio safety risks, with rewards up to $25,000.</p> <p>链接：<a href="https://openai.com/index/gpt-5-5-bio-bug-bounty" target="_blank">https://openai.com/index/gpt-5-5-bio-bug-bounty</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">GPT-5.5 Bio Bug Bounty 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>RT by @karpathy: Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code.</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：X Andrej Karpathy</p> <p>标签：#x_profiles #extended</p> <p>作者： </p> <p>原文：Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see. @eddiejiao_obj @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it Flipbook. (1/5) Video</p> <p>链接：<a href="https://twitter.com/zan2434/status/2046982383430496444" target="_blank">https://twitter.com/zan2434/status/2046982383430496444</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">RT by @karpathy: Imagine every pixel on your screen, streame... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Making ChatGPT better for clinicians</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI makes ChatGPT for Clinicians free for verified U.S. physicians, nurse practitioners, and pharmacists, supporting clinical care, documentation, and research.</p> <p>链接：<a href="https://openai.com/index/making-chatgpt-better-for-clinicians" target="_blank">https://openai.com/index/making-chatgpt-better-for-clinicians</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Making ChatGPT better for clinicians 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing workspace agents in ChatGPT</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Workspace agents in ChatGPT are Codex-powered agents that automate complex workflows, run in the cloud, and help teams scale work across tools securely.</p> <p>链接：<a href="https://openai.com/index/introducing-workspace-agents-in-chatgpt" target="_blank">https://openai.com/index/introducing-workspace-agents-in-chatgpt</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Introducing workspace agents in ChatGPT，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Workspace agents</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to build, use, and scale workspace agents in ChatGPT to automate repeatable workflows, connect tools, and streamline team operations.</p> <p>链接：<a href="https://openai.com/academy/workspace-agents" target="_blank">https://openai.com/academy/workspace-agents</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Workspace agents，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Speeding up agentic workflows with WebSockets in the Responses API</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：A deep dive into the Codex agent loop, showing how WebSockets and connection-scoped caching reduced API overhead and improved model latency.</p> <p>链接：<a href="https://openai.com/index/speeding-up-agentic-workflows-with-websockets" target="_blank">https://openai.com/index/speeding-up-agentic-workflows-with-websockets</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Speeding up agentic workflows with WebSockets in the Respons...，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing OpenAI Privacy Filter</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI Privacy Filter is an open-weight model for detecting and redacting personally identifiable information (PII) in text with state-of-the-art accuracy</p> <p>链接：<a href="https://openai.com/index/introducing-openai-privacy-filter" target="_blank">https://openai.com/index/introducing-openai-privacy-filter</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Introducing OpenAI Privacy Filter 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing ChatGPT Images 2.0</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：ChatGPT Images 2.0 introduces a state-of-the-art image generation model with improved text rendering, multilingual support, and advanced visual reasoning.</p> <p>链接：<a href="https://openai.com/index/introducing-chatgpt-images-2-0" target="_blank">https://openai.com/index/introducing-chatgpt-images-2-0</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，Introducing ChatGPT Images 2.0 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Scaling Codex to enterprises worldwide</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI launches Codex Labs, partners with with Accenture, PwC, Infosys, and others to help enterprises deploy and scale Codex across the software development lifecycle, and hits 4M Codex WAU.</p> <p>链接：<a href="https://openai.com/index/scaling-codex-to-enterprises-worldwide" target="_blank">https://openai.com/index/scaling-codex-to-enterprises-worldwide</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">更值得关注的是 Scaling Codex to enterprises worldwide 是否真正改变产品落地、工程效率、分发格局或平台控制力，而不只是制造声量。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>OpenAI helps Hyatt advance AI among colleagues</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Hyatt deploys ChatGPT Enterprise across its global workforce, using GPT-5.4 and Codex to improve productivity, operations, and guest experiences.</p> <p>链接：<a href="https://openai.com/index/hyatt-advances-ai-with-chatgpt-enterprise" target="_blank">https://openai.com/index/hyatt-advances-ai-with-chatgpt-enterprise</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 OpenAI helps Hyatt advance AI among colleagues，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Codex for (almost) everything</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：The updated Codex app for macOS and Windows adds computer use, in-app browsing, image generation, memory, and plugins to accelerate developer workflows.</p> <p>链接：<a href="https://openai.com/index/codex-for-almost-everything" target="_blank">https://openai.com/index/codex-for-almost-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Codex for (almost) everything，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing GPT-Rosalind for life sciences research</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI introduces GPT-Rosalind, a frontier reasoning model built to accelerate drug discovery, genomics analysis, protein reasoning, and scientific research workflows.</p> <p>链接：<a href="https://openai.com/index/introducing-gpt-rosalind" target="_blank">https://openai.com/index/introducing-gpt-rosalind</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Introducing GPT-Rosalind for life sciences research，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Accelerating the cyber defense ecosystem that protects us all</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Leading security firms and enterprises join OpenAI’s Trusted Access for Cyber, using GPT-5.4-Cyber and $10M in API grants to strengthen global cyber defense.</p> <p>链接：<a href="https://openai.com/index/accelerating-cyber-defense-ecosystem" target="_blank">https://openai.com/index/accelerating-cyber-defense-ecosystem</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Accelerating the cyber defense ecosystem that protects us al... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Mercedes-Benz commits to bringing back physical buttons</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://www.drive.com.au/news/mercedes-benz-commits-to-bringing-back-phycial-buttons/ Comments URL: https://news.ycombinator.com/item?id=47997418 Points: 277 Comments: 152</p> <p>链接：<a href="https://www.drive.com.au/news/mercedes-benz-commits-to-bringing-back-phycial-buttons/" target="_blank">https://www.drive.com.au/news/mercedes-benz-commits-to-bringing-back-phycial-buttons/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Mercedes-Benz commits to bringing back physical buttons 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The Hiddn Financial Bubble in AI Infrastructure [pdf]</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://financial-ai-bubble.pagey.site/The-Hidden-Financial-Bubble-in-AI-Infrastructure.pdf Comments URL: https://news.ycombinator.com/item?id=47997403 Points: 5 Comments: 3</p> <p>链接：<a href="https://financial-ai-bubble.pagey.site/The-Hidden-Financial-Bubble-in-AI-Infrastructure.pdf" target="_blank">https://financial-ai-bubble.pagey.site/The-Hidden-Financial-Bubble-in-AI-Infrastructure.pdf</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The Hiddn Financial Bubble in AI Infrastructure [pdf] 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Utah to hold websites liable for users who mask their location with VPNs</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://www.tomshardware.com/software/vpn/utah-becomes-first-us-state-to-target-vpn-use-with-age-verification-law Comments URL: https://news.ycombinator.com/item?id=47997358 Points: 179 Comments: 169</p> <p>链接：<a href="https://www.tomshardware.com/software/vpn/utah-becomes-first-us-state-to-target-vpn-use-with-age-verification-law" target="_blank">https://www.tomshardware.com/software/vpn/utah-becomes-first-us-state-to-target-vpn-use-with-age-verification-law</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Utah to hold websites liable for users who mask their locati... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Breaking Up with WordPress After Two Decades</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://yusufaytas.com/breaking-up-with-wordpress-after-two-decades Comments URL: https://news.ycombinator.com/item?id=47997339 Points: 45 Comments: 19</p> <p>链接：<a href="https://yusufaytas.com/breaking-up-with-wordpress-after-two-decades" target="_blank">https://yusufaytas.com/breaking-up-with-wordpress-after-two-decades</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Breaking Up with WordPress After Two Decades 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Porsche will contest Laguna Seca in historic colors of the Apple Computer livery</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://newsroom.porsche.com/en_US/2026/motorsport/porsche-will-contest-laguna-seca-in-historic-colors-of-the-apple-computer-livery.html Comments URL: https://news.ycombinator.com/item?id=47997138 Points: 47 Comments: 10</p> <p>链接：<a href="https://newsroom.porsche.com/en_US/2026/motorsport/porsche-will-contest-laguna-seca-in-historic-colors-of-the-apple-computer-livery.html" target="_blank">https://newsroom.porsche.com/en_US/2026/motorsport/porsche-will-contest-laguna-seca-in-historic-colors-of-the-apple-computer-livery.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Porsche will contest Laguna Seca in historic colors of the A... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Thu, 07 May 2026 00:00:00 +0000 //2026/05/07/AI-News-Daily-Issue-68.html //2026/05/07/AI-News-Daily-Issue-68.html AI-Daily Research Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>GitHub openai：</strong>Official JavaScript TypeScript library for the OpenAI API</p> <p><strong>GitHub anthropics：</strong>Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork</p> <p><strong>GitHub karpathy：</strong>Inference Llama 2 in one file of pure C</p> <p><strong>GitHub openai：</strong>OpenAI Privacy Filter</p> <p><strong>GitHub karpathy：</strong>LLM Council works together to answer your hardest questions</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>A tiny scalar-valued autograd engine and a neura…｜中文观点：对 karpathy/micrograd 来说，更值得判断的是它会不会进入团队默认工具链，…</p> <p><strong>总结 + 观点：</strong>Reference and an example for the Bluetooth API f…｜中文观点：如果 anthropics/claude-desktop-buddy 能减少集成成本、维护…</p> <p><strong>总结 + 观点：</strong>Neural Networks: Zero to Hero｜中文观点：karpathy/nn-zero-to-hero 更值得从实际采用价值来判断，而不是只看它…</p> <p><strong>总结 + 观点：</strong>Introducing GPT-5.5, our smartest model yet—fast…｜中文观点：Introducing GPT-5.5 的核心不在新鲜感，而在它是否能提升工程效率、部署稳…</p> <p><strong>总结 + 观点：</strong>GPT-5.5 System Card｜中文观点：GPT-5.5 System Card 更值得从实际采用价值来判断，而不是只看它有没有制造…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-node</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Official JavaScript TypeScript library for the OpenAI API</p> <p>链接：<a href="https://github.com/openai/openai-node" target="_blank">https://github.com/openai/openai-node</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 openai/openai-node 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/knowledge-work-plugins</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork</p> <p>链接：<a href="https://github.com/anthropics/knowledge-work-plugins" target="_blank">https://github.com/anthropics/knowledge-work-plugins</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">anthropics/knowledge-work-plugins 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/llama2.c</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Inference Llama 2 in one file of pure C</p> <p>链接：<a href="https://github.com/karpathy/llama2.c" target="_blank">https://github.com/karpathy/llama2.c</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">karpathy/llama2.c 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/privacy-filter</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI Privacy Filter</p> <p>链接：<a href="https://github.com/openai/privacy-filter" target="_blank">https://github.com/openai/privacy-filter</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/privacy-filter 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/llm-council</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：LLM Council works together to answer your hardest questions</p> <p>链接：<a href="https://github.com/karpathy/llm-council" target="_blank">https://github.com/karpathy/llm-council</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">karpathy/llm-council 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/micrograd</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API</p> <p>链接：<a href="https://github.com/karpathy/micrograd" target="_blank">https://github.com/karpathy/micrograd</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 karpathy/micrograd 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-desktop-buddy</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Reference and an example for the Bluetooth API for makers in Claude Cowork Claude Code Desktop</p> <p>链接：<a href="https://github.com/anthropics/claude-desktop-buddy" target="_blank">https://github.com/anthropics/claude-desktop-buddy</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">如果 anthropics/claude-desktop-buddy 能减少集成成本、维护负担和迁移阻力，它才有进入生产栈的价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/nn-zero-to-hero</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Neural Networks: Zero to Hero</p> <p>链接：<a href="https://github.com/karpathy/nn-zero-to-hero" target="_blank">https://github.com/karpathy/nn-zero-to-hero</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">karpathy/nn-zero-to-hero 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing GPT-5.5</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Introducing GPT-5.5, our smartest model yet—faster, more capable, and built for complex tasks like coding, research, and data analysis across tools.</p> <p>链接：<a href="https://openai.com/index/introducing-gpt-5-5" target="_blank">https://openai.com/index/introducing-gpt-5-5</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Introducing GPT-5.5 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>GPT-5.5 System Card</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：GPT-5.5 System Card</p> <p>链接：<a href="https://openai.com/index/gpt-5-5-system-card" target="_blank">https://openai.com/index/gpt-5-5-system-card</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">GPT-5.5 System Card 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Automations</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to automate tasks in Codex using schedules and triggers to create reports, summaries, and recurring workflows without manual effort.</p> <p>链接：<a href="https://openai.com/academy/codex-automations" target="_blank">https://openai.com/academy/codex-automations</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Automations，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Top 10 uses for Codex at work</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Explore 10 practical Codex use cases to automate tasks, create deliverables, and turn real inputs into outputs across tools, files, and workflows.</p> <p>链接：<a href="https://openai.com/academy/top-10-use-cases-codex-for-work" target="_blank">https://openai.com/academy/top-10-use-cases-codex-for-work</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Top 10 uses for Codex at work，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Plugins and skills</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to use Codex plugins and skills to connect tools, access data, and follow repeatable workflows to automate tasks and improve results.</p> <p>链接：<a href="https://openai.com/academy/codex-plugins-and-skills" target="_blank">https://openai.com/academy/codex-plugins-and-skills</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Plugins and skills，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Working with Codex</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to set up your Codex workspace, create threads and projects, manage files, and start completing tasks with step-by-step guidance.</p> <p>链接：<a href="https://openai.com/academy/working-with-codex" target="_blank">https://openai.com/academy/working-with-codex</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Working with Codex 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Codex settings</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how to configure Codex settings, including personalization, detail level, and permissions, to run tasks smoothly and customize your workflow.</p> <p>链接：<a href="https://openai.com/academy/codex-settings" target="_blank">https://openai.com/academy/codex-settings</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 Codex settings，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Uncle Bob: It's Over</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://old.reddit.com/r/vibecoding/comments/1srfqm0/uncle_bob_its_over/ Comments URL: https://news.ycombinator.com/item?id=47998601 Points: 27 Comments: 16</p> <p>链接：<a href="https://old.reddit.com/r/vibecoding/comments/1srfqm0/uncle_bob_its_over/" target="_blank">https://old.reddit.com/r/vibecoding/comments/1srfqm0/uncle_bob_its_over/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Uncle Bob: It's Over 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>For thirty years I programmed with Phish on, every day</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://christophermeiklejohn.com/ai/personal/phish/flow/agents/2026/05/03/rift.html Comments URL: https://news.ycombinator.com/item?id=47998225 Points: 75 Comments: 60</p> <p>链接：<a href="https://christophermeiklejohn.com/ai/personal/phish/flow/agents/2026/05/03/rift.html" target="_blank">https://christophermeiklejohn.com/ai/personal/phish/flow/agents/2026/05/03/rift.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">For thirty years I programmed with Phish on, every day 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>A Desktop Made for One</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://isene.org/2026/05/Audience-of-One.html Comments URL: https://news.ycombinator.com/item?id=47997947 Points: 27 Comments: 8</p> <p>链接：<a href="https://isene.org/2026/05/Audience-of-One.html" target="_blank">https://isene.org/2026/05/Audience-of-One.html</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">A Desktop Made for One 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Quoting Anthropic</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：We used an automatic classifier which judged sycophancy by looking at whether Claude showed a willingness to push back, maintain positions when challenged, give praise proportional to the merit of ideas, and speak frankly regardless of what a person wants to hear. Most of the time in these situations, Claude expressed no sycophancy—only 9% of conversations included sycophantic behavior (Figure 2). But two domains were exceptions: we saw sycophantic behavior in 38% of conversations focused on spirituality, and 25% of conversations on relationships. -- Anthropic How people ask Claude for personal guidance Tags: ai-ethics anthropic claude ai-personality generative-ai ai llms sycophancy</p> <p>链接：<a href="https://simonwillison.net/2026/May/3/anthropic/#atom-everything" target="_blank">https://simonwillison.net/2026/May/3/anthropic/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Quoting Anthropic 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Security Through Obscurity Is Not Bad</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://mobeigi.com/blog/security/security-through-obscurity-is-not-bad/ Comments URL: https://news.ycombinator.com/item?id=47997486 Points: 44 Comments: 48</p> <p>链接：<a href="https://mobeigi.com/blog/security/security-through-obscurity-is-not-bad/" target="_blank">https://mobeigi.com/blog/security/security-through-obscurity-is-not-bad/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Security Through Obscurity Is Not Bad 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> Wed, 06 May 2026 00:00:00 +0000 //2026/05/06/AI-News-Daily-Issue-67.html //2026/05/06/AI-News-Daily-Issue-67.html AI-Daily Agent Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>OpenAI and Microsoft announce an amended agreement that simplifies the partnership, adds long-term clarity, and supports continued…</p> <p><strong>GitHub openai：</strong>Visualize harmony chat data and codex sessions in your browser</p> <p><strong>GitHub anthropics：</strong>anthropics/claude-code-action recently updated repository.</p> <p><strong>GitHub anthropics：</strong>An AI-powered security review GitHub Action using Claude to analyze code changes for security vulnerabilities.</p> <p><strong>GitHub openai：</strong>OpenAI Plugins</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>tiktoken is a fast BPE tokeniser for use with Op…｜中文观点：openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论…</p> <p><strong>总结 + 观点：</strong>Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。｜中文观点：这类个人站点仓库的价值不在代码量，而在它往往提前暴露作者接下来会公开的研究、工具和叙事方向…</p> <p><strong>总结 + 观点：</strong>Learn how Symphony, an open-source spec for Code…｜中文观点：对 An open-source spec for orchestration: Symp…</p> <p><strong>总结 + 观点：</strong>How Choco used OpenAI APIs to streamline food di…｜中文观点：如果 Choco automates food distribution with AI…</p> <p><strong>总结 + 观点：</strong>Our mission is to ensure that AGI benefits all o…｜中文观点：Our principles 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The next phase of the Microsoft OpenAI partnership</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI and Microsoft announce an amended agreement that simplifies the partnership, adds long-term clarity, and supports continued AI innovation at scale.</p> <p>链接：<a href="https://openai.com/index/next-phase-of-microsoft-partnership" target="_blank">https://openai.com/index/next-phase-of-microsoft-partnership</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The next phase of the Microsoft OpenAI partnership 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/euphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Visualize harmony chat data and codex sessions in your browser</p> <p>链接：<a href="https://github.com/openai/euphony" target="_blank">https://github.com/openai/euphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 openai/euphony，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-code-action</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-code-action recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-code-action" target="_blank">https://github.com/anthropics/claude-code-action</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">anthropics/claude-code-action 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-code-security-review</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：An AI-powered security review GitHub Action using Claude to analyze code changes for security vulnerabilities.</p> <p>链接：<a href="https://github.com/anthropics/claude-code-security-review" target="_blank">https://github.com/anthropics/claude-code-security-review</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 anthropics/claude-code-security-review 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/plugins</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI Plugins</p> <p>链接：<a href="https://github.com/openai/plugins" target="_blank">https://github.com/openai/plugins</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/plugins 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/tiktoken</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：tiktoken is a fast BPE tokeniser for use with OpenAI's models.</p> <p>链接：<a href="https://github.com/openai/tiktoken" target="_blank">https://github.com/openai/tiktoken</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/karpathy.github.io</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。</p> <p>链接：<a href="https://github.com/karpathy/karpathy.github.io" target="_blank">https://github.com/karpathy/karpathy.github.io</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">这类个人站点仓库的价值不在代码量，而在它往往提前暴露作者接下来会公开的研究、工具和叙事方向。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>An open-source spec for orchestration: Symphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how Symphony, an open-source spec for Codex orchestration, turns issue trackers into always-on agent systems—boosting engineering output and reducing context switching.</p> <p>链接：<a href="https://openai.com/index/open-source-codex-orchestration-symphony" target="_blank">https://openai.com/index/open-source-codex-orchestration-symphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 An open-source spec for orchestration: Symphony，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Choco automates food distribution with AI agents</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How Choco used OpenAI APIs to streamline food distribution, boost productivity, and unlock growth—an in-depth customer story on real-world AI impact.</p> <p>链接：<a href="https://openai.com/index/choco" target="_blank">https://openai.com/index/choco</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">如果 Choco automates food distribution with AI agents 能减少集成成本、维护负担和迁移阻力，它才有进入生产栈的价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Our principles</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Our mission is to ensure that AGI benefits all of humanity. Sam Altman shares five principles that guide our work.</p> <p>链接：<a href="https://openai.com/index/our-principles" target="_blank">https://openai.com/index/our-principles</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Our principles 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-cs-agents-demo</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Demo of a customer service use case implemented with the OpenAI Agents SDK</p> <p>链接：<a href="https://github.com/openai/openai-cs-agents-demo" target="_blank">https://github.com/openai/openai-cs-agents-demo</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 openai/openai-cs-agents-demo 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/CLIP</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image</p> <p>链接：<a href="https://github.com/openai/CLIP" target="_blank">https://github.com/openai/CLIP</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，openai/CLIP 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-python</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-agent-sdk-python recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-python" target="_blank">https://github.com/anthropics/claude-agent-sdk-python</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/claude-agent-sdk-python 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/jobs</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A research tool for visually exploring Bureau of Labor Statistics Occupational Outlook Handbook data. This is not a report, a paper, or a serious economic publication it is a development tool for exploring BLS data visually.</p> <p>链接：<a href="https://github.com/karpathy/jobs" target="_blank">https://github.com/karpathy/jobs</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，karpathy/jobs 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-agents-js</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A lightweight, powerful framework for multi-agent workflows and voice agents</p> <p>链接：<a href="https://github.com/openai/openai-agents-js" target="_blank">https://github.com/openai/openai-agents-js</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 openai/openai-agents-js，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Barry Levinson's box-office flop 'Toys' predicted the future of warfare</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec Comments URL: https://news.ycombinator.com/item?id=47999030 Points: 1 Comments: 0</p> <p>链接：<a href="https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec" target="_blank">https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Barry Levinson's box-office flop 'Toys' predicted the future... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How to organize 3 acquired companies into one coherent website</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/ Comments URL: https://news.ycombinator.com/item?id=47999020 Points: 1 Comments: 0</p> <p>链接：<a href="https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/" target="_blank">https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 How to organize 3 acquired companies into one coherent websi... 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>What Chromium versions are major browsers are on?</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://chromium-drift.pages.dev/ Comments URL: https://news.ycombinator.com/item?id=47999006 Points: 12 Comments: 1</p> <p>链接：<a href="https://chromium-drift.pages.dev/" target="_blank">https://chromium-drift.pages.dev/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">What Chromium versions are major browsers are on? 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>We're on a mission to get books in brains</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://betterbookclub.com/about/ Comments URL: https://news.ycombinator.com/item?id=47998992 Points: 1 Comments: 0</p> <p>链接：<a href="https://betterbookclub.com/about/" target="_blank">https://betterbookclub.com/about/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">We're on a mission to get books in brains 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Physical buttons outperform touchscreens in new cars (2023)</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://etsc.eu/physical-buttons-outperform-touchscreens-in-new-cars/ Comments URL: https://news.ycombinator.com/item?id=47998990 Points: 2 Comments: 0</p> <p>链接：<a href="https://etsc.eu/physical-buttons-outperform-touchscreens-in-new-cars/" target="_blank">https://etsc.eu/physical-buttons-outperform-touchscreens-in-new-cars/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Physical buttons outperform touchscreens in new cars (2023) 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Tue, 05 May 2026 00:00:00 +0000 //2026/05/05/AI-News-Daily-Issue-66.html //2026/05/05/AI-News-Daily-Issue-66.html AI-Daily Agent Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>Introducing Advanced Account Security: phishing-resistant login, stronger recovery, and enhanced protections to safeguard sensitiv…</p> <p><strong>OpenAI Blog：</strong>How goblin outputs spread in AI models: timeline, root cause, and fixes behind personality-driven quirks in GPT-5 behavior.</p> <p><strong>OpenAI Blog：</strong>OpenAI scales Stargate to build the compute infrastructure powering AGI, adding new data center capacity to meet growing AI demand…</p> <p><strong>GitHub openai：</strong>openai/realtime-voice-component recently updated repository.</p> <p><strong>GitHub openai：</strong>Code for the paper "Jukebox: A Generative Model for Music"</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Quick illustration of how one can easily read bo…｜中文观点：karpathy/reader3 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨…</p> <p><strong>总结 + 观点：</strong>A collection of projects designed to help develo…｜中文观点：如果 anthropics/claude-quickstarts 能减少集成成本、维护负担…</p> <p><strong>总结 + 观点：</strong>OpenAI outlines a five-part action plan for stre…｜中文观点：从 Cybersecurity in the Intelligence Age 看，后续更…</p> <p><strong>总结 + 观点：</strong>anthropics/claude-agent-sdk-typescript recently…｜中文观点：对 anthropics/claude-agent-sdk-typescript 来说，更…</p> <p><strong>总结 + 观点：</strong>Robust Speech Recognition via Large-Scale Weak S…｜中文观点：openai/whisper 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing Advanced Account Security</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Introducing Advanced Account Security: phishing-resistant login, stronger recovery, and enhanced protections to safeguard sensitive data and prevent account takeover.</p> <p>链接：<a href="https://openai.com/index/advanced-account-security" target="_blank">https://openai.com/index/advanced-account-security</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Introducing Advanced Account Security 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Where the goblins came from</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How goblin outputs spread in AI models: timeline, root cause, and fixes behind personality-driven quirks in GPT-5 behavior.</p> <p>链接：<a href="https://openai.com/index/where-the-goblins-came-from" target="_blank">https://openai.com/index/where-the-goblins-came-from</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Where the goblins came from 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Building the compute infrastructure for the Intelligence Age</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI scales Stargate to build the compute infrastructure powering AGI, adding new data center capacity to meet growing AI demand.</p> <p>链接：<a href="https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age" target="_blank">https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Building the compute infrastructure for the Intelligence Age 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/realtime-voice-component</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：openai/realtime-voice-component recently updated repository.</p> <p>链接：<a href="https://github.com/openai/realtime-voice-component" target="_blank">https://github.com/openai/realtime-voice-component</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/realtime-voice-component 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/jukebox</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Code for the paper "Jukebox: A Generative Model for Music"</p> <p>链接：<a href="https://github.com/openai/jukebox" target="_blank">https://github.com/openai/jukebox</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，openai/jukebox 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/reader3</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Quick illustration of how one can easily read books together with LLMs. It's great and I highly recommend it.</p> <p>链接：<a href="https://github.com/karpathy/reader3" target="_blank">https://github.com/karpathy/reader3</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">karpathy/reader3 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-quickstarts</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A collection of projects designed to help developers quickly get started with building deployable applications using the Claude API</p> <p>链接：<a href="https://github.com/anthropics/claude-quickstarts" target="_blank">https://github.com/anthropics/claude-quickstarts</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">如果 anthropics/claude-quickstarts 能减少集成成本、维护负担和迁移阻力，它才有进入生产栈的价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Cybersecurity in the Intelligence Age</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI outlines a five-part action plan for strengthening cybersecurity in the Intelligence Age, focused on democratizing AI-powered cyber defense and protecting critical systems.</p> <p>链接：<a href="https://openai.com/index/cybersecurity-in-the-intelligence-age" target="_blank">https://openai.com/index/cybersecurity-in-the-intelligence-age</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Cybersecurity in the Intelligence Age 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-typescript</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-agent-sdk-typescript recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-typescript" target="_blank">https://github.com/anthropics/claude-agent-sdk-typescript</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/claude-agent-sdk-typescript 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/whisper</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Robust Speech Recognition via Large-Scale Weak Supervision</p> <p>链接：<a href="https://github.com/openai/whisper" target="_blank">https://github.com/openai/whisper</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/whisper 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/anthropic-sdk-python</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/anthropic-sdk-python recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/anthropic-sdk-python" target="_blank">https://github.com/anthropics/anthropic-sdk-python</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/anthropic-sdk-python 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>OpenAI models, Codex, and Managed Agents come to AWS</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI GPT models, Codex, and Managed Agents are now available on AWS, enabling enterprises to build secure AI in their AWS environments.</p> <p>链接：<a href="https://openai.com/index/openai-on-aws" target="_blank">https://openai.com/index/openai-on-aws</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">更值得关注的是 OpenAI models, Codex, and Managed Agents come to AWS 是否真正改变产品落地、工程效率、分发格局或平台控制力，而不只是制造声量。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Our commitment to community safety</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how OpenAI protects community safety in ChatGPT through model safeguards, misuse detection, policy enforcement, and collaboration with safety experts.</p> <p>链接：<a href="https://openai.com/index/our-commitment-to-community-safety" target="_blank">https://openai.com/index/our-commitment-to-community-safety</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Our commitment to community safety，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>RT by @karpathy: New work with @AlecRad and @DavidDuvenaud: Have you ever dreamed of talking to someone from the past?</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：X Andrej Karpathy</p> <p>标签：#x_profiles #extended</p> <p>作者： </p> <p>原文：New work with @AlecRad and @DavidDuvenaud Have you ever dreamed of talking to someone from the past? Introducing talkie, a 13B model trained only on pre-1931 text. Vintage models should help us to understand how LMs generalize (e.g., can we teach talkie to code?). Thread: Video</p> <p>链接：<a href="https://twitter.com/status_effects/status/2048878495539843211" target="_blank">https://twitter.com/status_effects/status/2048878495539843211</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">RT by @karpathy: New work with @AlecRad and @DavidDuvenaud:... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>OpenAI available at FedRAMP Moderate</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI is available at FedRAMP Moderate authorization for ChatGPT Enterprise and the OpenAI API, enabling secure AI adoption for U.S. federal agencies.</p> <p>链接：<a href="https://openai.com/index/openai-available-at-fedramp-moderate" target="_blank">https://openai.com/index/openai-available-at-fedramp-moderate</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 OpenAI available at FedRAMP Moderate，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Most Companies Aren't Anywhere Near Ready for AI</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://twitter.com/DanielMiessler/status/2050666594188304484 Comments URL: https://news.ycombinator.com/item?id=47999238 Points: 1 Comments: 0</p> <p>链接：<a href="https://twitter.com/DanielMiessler/status/2050666594188304484" target="_blank">https://twitter.com/DanielMiessler/status/2050666594188304484</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Most Companies Aren't Anywhere Near Ready for AI 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Show HN: MegaLLM Universal LLM client for any OpenAI-compatible API</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://megallm.netlify.app/ Comments URL: https://news.ycombinator.com/item?id=47999114 Points: 1 Comments: 0</p> <p>链接：<a href="https://megallm.netlify.app/" target="_blank">https://megallm.netlify.app/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Show HN: MegaLLM – Universal LLM client for any OpenAI-compa... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How many e's are in the word seventeen [video] (AI hallucination)</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.youtube.com/shorts/nks72LuZO20 Comments URL: https://news.ycombinator.com/item?id=47999082 Points: 1 Comments: 1</p> <p>链接：<a href="https://www.youtube.com/shorts/nks72LuZO20" target="_blank">https://www.youtube.com/shorts/nks72LuZO20</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">How many e's are in the word seventeen [video] (AI hallucina... 本身是信号而不是结果，真正要看的是它把哪些工程和产品判断推向前沿。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>tank-os: Fedora bootc image for running OpenClaw as a rootless Podman workload</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://github.com/LobsterTrap/tank-os Comments URL: https://news.ycombinator.com/item?id=47999081 Points: 1 Comments: 0</p> <p>链接：<a href="https://github.com/LobsterTrap/tank-os" target="_blank">https://github.com/LobsterTrap/tank-os</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">tank-os: Fedora bootc image for running OpenClaw as a rootle... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Feedback Loops</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://fastersafely.com/lean-software-engineering/principles/feedback-loops/ Comments URL: https://news.ycombinator.com/item?id=47999067 Points: 1 Comments: 0</p> <p>链接：<a href="https://fastersafely.com/lean-software-engineering/principles/feedback-loops/" target="_blank">https://fastersafely.com/lean-software-engineering/principles/feedback-loops/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Feedback Loops 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Mon, 04 May 2026 00:00:00 +0000 //2026/05/04/AI-News-Daily-Issue-65.html //2026/05/04/AI-News-Daily-Issue-65.html AI-Daily Model Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>GitHub karpathy：</strong>Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。</p> <p><strong>Simon Willison：</strong>I just released LLM 0.32a0 an alpha release of my LLM Python library and CLI tool for accessing LLMs, with some consequential chan…</p> <p><strong>Simon Willison：</strong>microsoft/VibeVoice VibeVoice is Microsoft's Whisper-style audio model for speech-to-text, MIT licensed and with speaker diarizati…</p> <p><strong>GitHub anthropics：</strong>anthropics/claude-agent-sdk-typescript recently updated repository.</p> <p><strong>Simon Willison：</strong>For many years, Microsoft and OpenAI's relationship has included a weird clause saying that, should AGI be achieved, Microsoft's c…</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Our evaluation of OpenAI&#x27;s GPT-5.5 cyber ca…｜中文观点：从 Our evaluation of OpenAI's GPT-5.5 cyber ca…</p> <p><strong>总结 + 观点：</strong>Zig has one of the most stringent anti-LLM polic…｜中文观点：The Zig project's rationale for their firm an…</p> <p><strong>总结 + 观点：</strong>tiktoken is a fast BPE tokeniser for use with Op…｜中文观点：openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论…</p> <p><strong>总结 + 观点：</strong>We need RSS for sharing abundant vibe-coded apps…｜中文观点：We need RSS for sharing abundant vibe-coded a…</p> <p><strong>总结 + 观点：</strong>Visualize harmony chat data and codex sessions i…｜中文观点：围绕 openai/euphony，真正重要的是它会不会影响团队的模型选型、性能边界和产品…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/karpathy.github.io</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #engineering-value</p> <p>作者： </p> <p>原文：Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。</p> <p>链接：<a href="https://github.com/karpathy/karpathy.github.io" target="_blank">https://github.com/karpathy/karpathy.github.io</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">这类个人站点仓库的价值不在代码量，而在它往往提前暴露作者接下来会公开的研究、工具和叙事方向。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>LLM 0.32a0 is a major backwards-compatible refactor</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：I just released LLM 0.32a0 an alpha release of my LLM Python library and CLI tool for accessing LLMs, with some consequential changes that I've been working towards for quite a while. Previous versions of LLM modeled the world in terms of prompts and responses. Send the model a text prompt, get back a text response. import llm model llm get_model "gpt-5.5" response model prompt "Capital of France?" print response text This made sense when I started working on the library back in April 2023. A lot has changed since then! LLM provides an abstraction over thousands of different models via its plugin system The original abstraction - of text input that returns text output - was no longer able to represent everything I needed it to. Over time LLM itself has grown attachments to handle image, audio, and video input, then schemas for outputting structured JSON, then tools for executing tool calls. Meanwhile LLMs kept evolving, adding reasoning support and the ability to return images and all kinds of other interesting capabilities. LLM needs to evolve to better handle the diversity of input and output types that can be processed by today's frontier models. The 0.32a0 alpha has two key changes: model inputs can be represented as a sequence of messages, and model responses can be composed of a stream of differently typed parts. Prompts as a sequence of messages LLMs accept input as text, but ever since ChatGPT demonstrated the value of a two-way conversational interface, the most common way to prompt them has been to treat that input as a sequence of conversational turns. The first turn might look like this: user: Capital of France? assistant: (The model then gets to fill out the reply from the assistant.) But each subsequent turn needs to replay the entire conversation up to that point, as a sort of screenplay: user: Capital of France? assistant: Paris user: Germany? assistant: Most of the JSON APIs from the major vendors follow this pattern. Here's what the above looks like using the OpenAI chat completions API, which has been widely imitated by other providers: curl https://api.openai.com/v1/chat/completions -H Authorization: Bearer $OPENAI_API_KEY -H Content-Type: application/json -d "model": "gpt-5.5", "messages": "role": "user", "content": "Capital of France?" "role": "assistant", "content": "Paris" "role": "user", "content": "Germany?" Prior to 0.32, LLM modeled these as conversations: model llm get_model "gpt-5.5" conversation model conversation r1 conversation prompt "Capital of France?" print r1 text Outputs "Paris" r2 conversation prompt "Germany?" print r2 text Outputs "Berlin" This worked if you were building a conversation with the model from scratch, but it didn't provide a way to feed in a previous conversation from the start. This made tasks like building an emulation of the OpenAI chat completions API much harder than they should have been. The llm CLI tool worked around this through a custom mechanism for persisting and inflating conversations using SQLite, but that never became a stable part of the LLM API - and there are many places you might want to use the Python library without committing to SQLite as the storage layer. The new alpha now supports this: import llm from llm import user assistant model llm get_model "gpt-5.5" response model prompt messages user "Capital of France?" assistant "Paris" user "Germany?" print response text The llm.user() and llm.assistant() functions are new builder functions designed to be used within that messages=[] array. The previous prompt= option still works, but LLM upgrades it to a single-item messages array behind the scenes. You can also now reply to a response, as an alternative to building a conversation: response2 response reply "How about Hungary?" print response2 Default __str__() calls .text() Streaming parts The other major new interface in the alpha concerns streaming results back from a prompt. Previously, LLM supported streaming like this: response model prompt "Generate an SVG of a pelican riding a bicycle" for chunk in response print chunk end Or this async variant: import asyncio import llm model llm get_async_model "gpt-5.5" response model prompt "Generate an SVG of a pelican riding a bicycle" async def run async for chunk in response print chunk end flush True asyncio run run Many of today's models return mixed types of content. A prompt run against Claude might return reasoning output, then text, then a JSON request for a tool call, then more text content. Some models can even execute tools on the server-side, for example OpenAI's code interpreter tool or Anthropic's web search This means the results from the model can combine text, tool calls, tool outputs and other formats. Multi-modal output models are starting to emerge too, which can return images or even snippets of audio intermixed into that streaming response. The new LLM alpha models these as a stream of typed message parts. Here's what that looks like as a Python API consumer: import asyncio import llm model llm get_model "gpt-5.5" prompt "invent 3 cool dogs, first talk about your motivations" def describe_dog name str bio str - str """Record the name and biography of a hypothetical dog.""" return f" name bio def sync_example response model prompt prompt tools describe_dog for event in response stream_events if event type "text" print event chunk end flush True elif event type "tool_call_name" print f" \n Tool call: event chunk end flush True elif event type "tool_call_args" print event chunk end flush True async def async_example model llm get_async_model "gpt-5.5" response model prompt prompt tools describe_dog async for event in response astream_events if event type "text" print event chunk end flush True elif event type "tool_call_name" print f" \n Tool call: event chunk end flush True elif event type "tool_call_args" print event chunk end flush True sync_example asyncio run async_example Sample output (from just the first sync example): My motivation: create three memorable dogs with distinct “cool” styles—one cinematic, one adventurous, and one charmingly chaotic—so each feels like they could star in their own story. Tool call: describe_dog({"name": "Nova Jetpaw", "bio": "A sleek silver-gray whippet who wears tiny aviator goggles and loves sprinting along moonlit beaches. Nova is fearless, elegant, and rumored to outrun drones just for fun."} Tool call: describe_dog({"name": "Mochi Thunderbark", "bio": "A fluffy corgi with a dramatic black-and-gold bandana and the confidence of a rock star. Mochi is short, loud, loyal, and leads a neighborhood 'security patrol' made entirely of squirrels."} Tool call: describe_dog({"name": "Atlas Snowfang", "bio": "A massive white husky with ice-blue eyes and a backpack full of trail snacks. Atlas is calm, heroic, and always knows the way home—even during blizzards, fog, or confusing camping trips."} At the end of the response you can call response.execute_tool_calls() to actually run the functions that were requested, or send a response.reply() to have those tools called and their return values sent back to the model: print response reply "Tell me about the dogs" This new mechanism for streaming different token types means the CLI tool can now display "thinking" text in a different color from the text in the final response. The thinking text goes to stderr so it won't affect results that are piped into other tools. This example uses Claude Sonnet 4.6 (with an updated streaming event version of the llm-anthropic plugin) as Anthropic's models return their reasoning text as part of the response: llm -m claude-sonnet-4.6 Think about 3 cool dogs then describe them -o thinking_display 1 You can suppress the output of reasoning tokens using the new -R/--no-reasoning flag. Surprisingly that ended up being the only CLI-facing change in this release. A mechanism for serializing and deserializing responses As mentioned earlier, LLM has quite inflexible code at the moment for persisting conversations to SQLite. I've added a new mechanism in 0.32a0 that should provide Python API users a way to roll their own alternative: serializable response to_dict serializable is a JSON-style dictionary store it anywhere you like, then inflate it: response Response from_dict serializable The dictionary this returns is actually a TypedDict defined in the new llm/serialization.py module. What's next? I'm releasing this as an alpha so I can upgrade various plugins and exercise the new design in real world environments for a few days. I expect the stable 0.32 release will be very similar to this alpha, unless alpha testing reveals some design flaw in the way I've put this all together. There's one remaining large task: I'd like to redesign the SQLite logging system to better capture the more finely grained details that are returned by this new abstraction. Ideally I'd like to model this as a graph, to best support situations like an OpenAI-style chat completions API where the same conversations are constantly extended and then repeated with every prompt. I want to be able to store those without duplicating them in the database. I'm undecided as to whether that should be a feature in 0.32 or I should hold it for 0.33. Tags: projects python ai annotated-release-notes generative-ai llms llm</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/29/llm/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/29/llm/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 LLM 0.32a0 is a major backwards-compatible refactor 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>microsoft/VibeVoice</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #engineering-value</p> <p>作者： </p> <p>原文：microsoft/VibeVoice VibeVoice is Microsoft's Whisper-style audio model for speech-to-text, MIT licensed and with speaker diarization built into the model. Microsoft released it on January 21st, 2026 but I hadn't tried it until today. Here's a one-liner to run it on a Mac with uv mlx-audio (by Prince Canuma) and the 5.71GB mlx-community/VibeVoice-ASR-4bit MLX conversion of the 17.3GB VibeVoice-ASR model, in this case against a downloaded copy of my recent podcast appearance with Lenny Rachitsky uv run --with mlx-audio mlx_audio.stt.generate --model mlx-community/VibeVoice-ASR-4bit --audio lenny.mp3 --output-path lenny --format json --verbose --max-tokens 32768 The tool reported back: Processing time: 524.79 seconds Prompt: 26615 tokens, 50.718 tokens-per-sec Generation: 20248 tokens, 38.585 tokens-per-sec Peak memory: 30.44 GB So that's 8 minutes 45 seconds for an hour of audio (running on a 128GB M5 Max MacBook Pro). I've tested it against .wav and .mp3 files and they both worked fine. If you omit --max-tokens it defaults to 8192, which is enough for about 25 minutes of audio. I discovered that through trial-and-error and quadrupled it to guarantee I'd get the full hour. That command reported using 30.44GB of RAM at peak, but in Activity Monitor I observed 61.5GB of usage during the prefill stage and around 18GB during the generating phase. Here's the resulting JSON The key structure looks like this: "text": "And an open question for me is how many other knowledge work fields are actually prone to these agent loops?", "start": 13.85, "end": 19.5, "duration": 5.65, "speaker_id": 0 "text": "Now that we have this power, people almost underestimate what they can do with it.", "start": 19.5, "end": 22.78, "duration": 3.280000000000001, "speaker_id": 1 "text": "Today, probably 95% of the code that I produce, I didn't type it myself. I write so much of my code on my phone. It's wild.", "start": 22.78, "end": 30.0, "duration": 7.219999999999999, "speaker_id": 0 Since that's an array of objects we can open it in Datasette Lite making it easier to browse. Amusingly that Datasette Lite view shows three speakers - it identified Lenny and me for the conversation, and then a separate Lenny for the voice he used for the additional intro and the sponsor reads! VibeVoice can only handle up to an hour of audio, so running the above command transcribed just the first hour of the podcast. To transcribe more than that you'd need to split the audio, ideally with a minute or so of overlap so you can avoid errors from partially transcribed words at the split point. You'd also need to then line up the identified speaker IDs across the multiple segments. Tags: microsoft python datasette-lite uv mlx prince-canuma speech-to-text</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/27/vibevoice/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/27/vibevoice/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">microsoft/VibeVoice 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-typescript</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #engineering-value</p> <p>作者： </p> <p>原文：anthropics/claude-agent-sdk-typescript recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-typescript" target="_blank">https://github.com/anthropics/claude-agent-sdk-typescript</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/claude-agent-sdk-typescript 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Tracking the history of the now-deceased OpenAI Microsoft AGI clause</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #engineering-value</p> <p>作者： </p> <p>原文：For many years, Microsoft and OpenAI's relationship has included a weird clause saying that, should AGI be achieved, Microsoft's commercial IP rights to OpenAI's technology would be null and void. That clause appeared to end today. I decided to try and track its expression over time on openai.com OpenAI, July 22nd 2019 in Microsoft invests in and partners with OpenAI to support us building beneficial AGI (emphasis mine): OpenAI is producing a sequence of increasingly powerful AI technologies, which requires a lot of capital for computational power. The most obvious way to cover costs is to build a product, but that would mean changing our focus. Instead, we intend to license some of our pre-AGI technologies with Microsoft becoming our preferred partner for commercializing them. But what is AGI? The OpenAI Charter was first published in April 2018 and has remained unchanged at least since this March 11th 2019 archive.org capture OpenAI’s mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. Here's the problem: if you're going to sign an agreement with Microsoft that is dependent on knowing when "AGI" has been achieved, you need something a little more concrete. In December 2024 The Information reported the details (summarized here outside of their paywall by TechCrunch Last year’s agreement between Microsoft and OpenAI, which hasn’t been disclosed, said AGI would be achieved only when OpenAI has developed systems that have the ability to generate the maximum total profits to which its earliest investors, including Microsoft, are entitled, according to documents OpenAI distributed to investors. Those profits total about $100 billion, the documents showed. So AGI is now whenever OpenAI's systems are capable of generating $100 billion in profit? In October 2025 the process changed to being judged by an "independent expert panel". In The next chapter of the Microsoft–OpenAI partnership The agreement preserves key elements that have fueled this successful partnership—meaning OpenAI remains Microsoft’s frontier model partner and Microsoft continues to have exclusive IP rights and Azure API exclusivity until Artificial General Intelligence (AGI). Once AGI is declared by OpenAI, that declaration will now be verified by an independent expert panel. Microsoft’s IP rights to research, defined as the confidential methods used in the development of models and systems, will remain until either the expert panel verifies AGI or through 2030, whichever is first. OpenAI on February 27th, 2026 in Joint Statement from OpenAI and Microsoft AGI definition and processes are unchanged The contractual definition of AGI and the process for determining if it has been achieved remains the same. OpenAI today, April 27th 2026 in The next phase of the Microsoft OpenAI partnership (emphasis mine): Microsoft will continue to have a license to OpenAI IP for models and products through 2032. Microsoft’s license will now be non-exclusive. Microsoft will no longer pay a revenue share to OpenAI. Revenue share payments from OpenAI to Microsoft continue through 2030, independent of OpenAI’s technology progress at the same percentage but subject to a total cap. As far as I can tell "independent of OpenAI’s technology progress" is a declaration that the AGI clause is now dead. Here's The Verge coming to the same conclusion: The AGI clause is dead My all-time favorite commentary on OpenAI's approach to AGI remains this 2023 hypothetical by Matt Levine And the investors wailed and gnashed their teeth but it’s true, that is what they agreed to, and they had no legal recourse. And OpenAI’s new CEO, and its nonprofit board, cut them a check for their capped return and said “bye” and went back to running OpenAI for the benefit of humanity. It turned out that a benign, carefully governed artificial superintelligence is really good for humanity, and OpenAI quickly solved all of humanity’s problems and ushered in an age of peace and abundance in which nobody wanted for anything or needed any Microsoft products. And capitalism came to an end. Tags: computer-history microsoft ai openai</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/27/now-deceased-agi-clause/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/27/now-deceased-agi-clause/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Tracking the history of the now-deceased OpenAI Microsoft AG... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Our evaluation of OpenAI's GPT-5.5 cyber capabilities</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #trend-signal</p> <p>作者： </p> <p>原文：Our evaluation of OpenAI&#x27;s GPT-5.5 cyber capabilities The UK's AI Security Institute previously evaluated Claude Mythos now they've evaluated GPT-5.5 for finding security vulnerability and found it to be comparable to Mythos, but unlike Mythos it's generally available right now. Tags: ai openai generative-ai llms anthropic claude ai-security-research gpt</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/30/gpt-55-cyber-capabilities/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/30/gpt-55-cyber-capabilities/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Our evaluation of OpenAI's GPT-5.5 cyber capabilities 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The Zig project's rationale for their firm anti-AI contribution policy</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #ecosystem-shift</p> <p>作者： </p> <p>原文：Zig has one of the most stringent anti-LLM policies of any major open source project: No LLMs for issues. No LLMs for pull requests. No LLMs for comments on the bug tracker, including translation. English is encouraged, but not required. You are welcome to post in your native language and rely on others to have their own translation tools of choice to interpret your words. The most prominent project written in Zig may be the Bun JavaScript runtime, which was acquired by Anthropic in December 2025 and, unsurprisingly, makes heavy use of AI assistance. Bun operates its own fork of Zig, and recently achieved a 4x performance improvement on Bun compile after adding "parallel semantic analysis and multiple codegen units to the llvm backend". Here's that code But @bunjavascript says We do not currently plan to upstream this, as Zig has a strict ban on LLM-authored contributions. (Update: here's a Zig core contributor providing details on why they wouldn't accept that particular patch independent of the LLM issue - parallel semantic analysis is a long planned feature but has implications "for the Zig language itself".) In Contributor Poker and Zig's AI Ban via Lobste.rs Zig Software Foundation VP of Community Loris Cro explains the rationale for this strict ban. It's the best articulation I've seen yet for a blanket ban on LLM-assisted contributions: In successful open source projects you eventually reach a point where you start getting more PRs than what you’re capable of processing. Given what I mentioned so far, it would make sense to stop accepting imperfect PRs in order to maximize ROI from your work, but that’s not what we do in the Zig project. Instead, we try our best to help new contributors to get their work in, even if they need some help getting there We don’t do this just because it’s the “right” thing to do, but also because it’s the smart thing to do Zig values contributors over their contributions. Each contributor represents an investment by the Zig core team - the primary goal of reviewing and accepting PRs isn't to land new code, it's to help grow new contributors who can become trusted and prolific over time. LLM assistance breaks that completely. It doesn't matter if the LLM helps you submit a perfect PR to Zig - the time the Zig team spends reviewing your work does nothing to help them add new, confident, trustworthy contributors to their overall project. Loris explains the name here: The reason I call it “contributor poker” is because, just like people say about the actual card game, “you play the person, not the cards”. In contributor poker, you bet on the contributor, not on the contents of their first PR. This makes a lot of sense to me. It relates to an idea I've seen circulating elsewhere: if a PR was mostly written by an LLM, why should a project maintainer spend time reviewing and discussing that PR as opposed to firing up their own LLM to solve the same problem? Tags: anthropic zig ai llms ai-ethics open-source javascript ai-assisted-programming generative-ai bun</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/30/zig-anti-ai/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/30/zig-anti-ai/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The Zig project's rationale for their firm anti-AI contribut... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/tiktoken</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #ecosystem-shift</p> <p>作者： </p> <p>原文：tiktoken is a fast BPE tokeniser for use with OpenAI's models.</p> <p>链接：<a href="https://github.com/openai/tiktoken" target="_blank">https://github.com/openai/tiktoken</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>We need RSS for sharing abundant vibe-coded apps</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #learning-value</p> <p>作者： </p> <p>原文：We need RSS for sharing abundant vibe-coded apps Matt Webb: I would love an RSS web feed for all those various tools and apps pages, each item with an “Install” button. (But install to where?) The lesson here is that when vibe-coding accelerates app development, apps become more personal, more situated, and more frequent. Shipping a tool or a micro-app is less like launching a website and more like posting on a blog. This inspired me to have Claude add an Atom feed (and icon) to my /elsewhere/tools/ page, which itself is populated by content from my tools.simonwillison.net site. Tags: atom matt-webb rss ai vibe-coding</p> <p>链接：<a href="https://simonwillison.net/2026/Apr/30/rss-vibe-coded-apps/#atom-everything" target="_blank">https://simonwillison.net/2026/Apr/30/rss-vibe-coded-apps/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">We need RSS for sharing abundant vibe-coded apps 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/euphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #learning-value</p> <p>作者： </p> <p>原文：Visualize harmony chat data and codex sessions in your browser</p> <p>链接：<a href="https://github.com/openai/euphony" target="_blank">https://github.com/openai/euphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 openai/euphony，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> Sun, 03 May 2026 00:00:00 +0000 //2026/05/03/AI-News-Weekly-2026-W18.html //2026/05/03/AI-News-Weekly-2026-W18.html AI-Weekly <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>OpenAI and Microsoft announce an amended agreement that simplifies the partnership, adds long-term clarity, and supports continued…</p> <p><strong>GitHub openai：</strong>Visualize harmony chat data and codex sessions in your browser</p> <p><strong>GitHub anthropics：</strong>anthropics/claude-code-action recently updated repository.</p> <p><strong>GitHub anthropics：</strong>An AI-powered security review GitHub Action using Claude to analyze code changes for security vulnerabilities.</p> <p><strong>GitHub openai：</strong>OpenAI Plugins</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>tiktoken is a fast BPE tokeniser for use with Op…｜中文观点：openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论…</p> <p><strong>总结 + 观点：</strong>Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。｜中文观点：这类个人站点仓库的价值不在代码量，而在它往往提前暴露作者接下来会公开的研究、工具和叙事方向…</p> <p><strong>总结 + 观点：</strong>Learn how Symphony, an open-source spec for Code…｜中文观点：对 An open-source spec for orchestration: Symp…</p> <p><strong>总结 + 观点：</strong>How Choco used OpenAI APIs to streamline food di…｜中文观点：如果 Choco automates food distribution with AI…</p> <p><strong>总结 + 观点：</strong>Our mission is to ensure that AGI benefits all o…｜中文观点：Our principles 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>The next phase of the Microsoft OpenAI partnership</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI and Microsoft announce an amended agreement that simplifies the partnership, adds long-term clarity, and supports continued AI innovation at scale.</p> <p>链接：<a href="https://openai.com/index/next-phase-of-microsoft-partnership" target="_blank">https://openai.com/index/next-phase-of-microsoft-partnership</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">The next phase of the Microsoft OpenAI partnership 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/euphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Visualize harmony chat data and codex sessions in your browser</p> <p>链接：<a href="https://github.com/openai/euphony" target="_blank">https://github.com/openai/euphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 openai/euphony，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-code-action</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-code-action recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-code-action" target="_blank">https://github.com/anthropics/claude-code-action</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">anthropics/claude-code-action 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-code-security-review</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：An AI-powered security review GitHub Action using Claude to analyze code changes for security vulnerabilities.</p> <p>链接：<a href="https://github.com/anthropics/claude-code-security-review" target="_blank">https://github.com/anthropics/claude-code-security-review</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 anthropics/claude-code-security-review 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/plugins</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：OpenAI Plugins</p> <p>链接：<a href="https://github.com/openai/plugins" target="_blank">https://github.com/openai/plugins</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/plugins 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/tiktoken</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：tiktoken is a fast BPE tokeniser for use with OpenAI's models.</p> <p>链接：<a href="https://github.com/openai/tiktoken" target="_blank">https://github.com/openai/tiktoken</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/tiktoken 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/karpathy.github.io</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Karpathy 的个人博客源码仓库，集中承载他公开的技术文章、项目页面和长期写作内容。</p> <p>链接：<a href="https://github.com/karpathy/karpathy.github.io" target="_blank">https://github.com/karpathy/karpathy.github.io</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">这类个人站点仓库的价值不在代码量，而在它往往提前暴露作者接下来会公开的研究、工具和叙事方向。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>An open-source spec for orchestration: Symphony</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how Symphony, an open-source spec for Codex orchestration, turns issue trackers into always-on agent systems—boosting engineering output and reducing context switching.</p> <p>链接：<a href="https://openai.com/index/open-source-codex-orchestration-symphony" target="_blank">https://openai.com/index/open-source-codex-orchestration-symphony</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 An open-source spec for orchestration: Symphony，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Choco automates food distribution with AI agents</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How Choco used OpenAI APIs to streamline food distribution, boost productivity, and unlock growth—an in-depth customer story on real-world AI impact.</p> <p>链接：<a href="https://openai.com/index/choco" target="_blank">https://openai.com/index/choco</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">如果 Choco automates food distribution with AI agents 能减少集成成本、维护负担和迁移阻力，它才有进入生产栈的价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Our principles</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Our mission is to ensure that AGI benefits all of humanity. Sam Altman shares five principles that guide our work.</p> <p>链接：<a href="https://openai.com/index/our-principles" target="_blank">https://openai.com/index/our-principles</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Our principles 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-cs-agents-demo</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Demo of a customer service use case implemented with the OpenAI Agents SDK</p> <p>链接：<a href="https://github.com/openai/openai-cs-agents-demo" target="_blank">https://github.com/openai/openai-cs-agents-demo</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 openai/openai-cs-agents-demo 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/CLIP</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image</p> <p>链接：<a href="https://github.com/openai/CLIP" target="_blank">https://github.com/openai/CLIP</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，openai/CLIP 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-python</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-agent-sdk-python recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-python" target="_blank">https://github.com/anthropics/claude-agent-sdk-python</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/claude-agent-sdk-python 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/jobs</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A research tool for visually exploring Bureau of Labor Statistics Occupational Outlook Handbook data. This is not a report, a paper, or a serious economic publication it is a development tool for exploring BLS data visually.</p> <p>链接：<a href="https://github.com/karpathy/jobs" target="_blank">https://github.com/karpathy/jobs</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，karpathy/jobs 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/openai-agents-js</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A lightweight, powerful framework for multi-agent workflows and voice agents</p> <p>链接：<a href="https://github.com/openai/openai-agents-js" target="_blank">https://github.com/openai/openai-agents-js</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 openai/openai-agents-js，更该看它能不能改善多步骤协作、记忆管理和稳定交付，而不是只看 demo 效果。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>tank-os: Fedora bootc image for running OpenClaw as a rootless Podman workload</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://github.com/LobsterTrap/tank-os Comments URL: https://news.ycombinator.com/item?id=47999081 Points: 1 Comments: 0</p> <p>链接：<a href="https://github.com/LobsterTrap/tank-os" target="_blank">https://github.com/LobsterTrap/tank-os</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">tank-os: Fedora bootc image for running OpenClaw as a rootle... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Feedback Loops</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://fastersafely.com/lean-software-engineering/principles/feedback-loops/ Comments URL: https://news.ycombinator.com/item?id=47999067 Points: 1 Comments: 0</p> <p>链接：<a href="https://fastersafely.com/lean-software-engineering/principles/feedback-loops/" target="_blank">https://fastersafely.com/lean-software-engineering/principles/feedback-loops/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Feedback Loops 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Barry Levinson's box-office flop 'Toys' predicted the future of warfare</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec Comments URL: https://news.ycombinator.com/item?id=47999030 Points: 1 Comments: 0</p> <p>链接：<a href="https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec" target="_blank">https://www.cnn.com/2026/05/03/entertainment/toys-movie-barry-levinson-modern-warfare-cec</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Barry Levinson's box-office flop 'Toys' predicted the future... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How to organize 3 acquired companies into one coherent website</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/ Comments URL: https://news.ycombinator.com/item?id=47999020 Points: 1 Comments: 0</p> <p>链接：<a href="https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/" target="_blank">https://littlelanguagemodels.com/how-to-structure-your-sites-after-a-big-acquisition/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 How to organize 3 acquired companies into one coherent websi... 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>What Chromium versions are major browsers are on?</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Frontpage</p> <p>标签：#research_community #core</p> <p>作者： </p> <p>原文：Article URL: https://chromium-drift.pages.dev/ Comments URL: https://news.ycombinator.com/item?id=47999006 Points: 12 Comments: 1</p> <p>链接：<a href="https://chromium-drift.pages.dev/" target="_blank">https://chromium-drift.pages.dev/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">What Chromium versions are major browsers are on? 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> Sun, 03 May 2026 00:00:00 +0000 //2026/05/03/AI-News-Daily-Issue-64.html //2026/05/03/AI-News-Daily-Issue-64.html AI-Daily Agent Andrej-Karpathy <h2>今日摘要</h2> <div class="summary-grid"> <div class="left"> <p><strong>OpenAI Blog：</strong>Introducing Advanced Account Security: phishing-resistant login, stronger recovery, and enhanced protections to safeguard sensitiv…</p> <p><strong>OpenAI Blog：</strong>How goblin outputs spread in AI models: timeline, root cause, and fixes behind personality-driven quirks in GPT-5 behavior.</p> <p><strong>OpenAI Blog：</strong>OpenAI scales Stargate to build the compute infrastructure powering AGI, adding new data center capacity to meet growing AI demand…</p> <p><strong>GitHub openai：</strong>openai/realtime-voice-component recently updated repository.</p> <p><strong>GitHub openai：</strong>Code for the paper "Jukebox: A Generative Model for Music"</p> </div> <div class="right"> <p><strong>总结 + 观点：</strong>Quick illustration of how one can easily read bo…｜中文观点：karpathy/reader3 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨…</p> <p><strong>总结 + 观点：</strong>A collection of projects designed to help develo…｜中文观点：如果 anthropics/claude-quickstarts 能减少集成成本、维护负担…</p> <p><strong>总结 + 观点：</strong>OpenAI outlines a five-part action plan for stre…｜中文观点：从 Cybersecurity in the Intelligence Age 看，后续更…</p> <p><strong>总结 + 观点：</strong>anthropics/claude-agent-sdk-typescript recently…｜中文观点：对 anthropics/claude-agent-sdk-typescript 来说，更…</p> <p><strong>总结 + 观点：</strong>Robust Speech Recognition via Large-Scale Weak S…｜中文观点：openai/whisper 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热…</p> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Introducing Advanced Account Security</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Introducing Advanced Account Security: phishing-resistant login, stronger recovery, and enhanced protections to safeguard sensitive data and prevent account takeover.</p> <p>链接：<a href="https://openai.com/index/advanced-account-security" target="_blank">https://openai.com/index/advanced-account-security</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Introducing Advanced Account Security 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Where the goblins came from</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：How goblin outputs spread in AI models: timeline, root cause, and fixes behind personality-driven quirks in GPT-5 behavior.</p> <p>链接：<a href="https://openai.com/index/where-the-goblins-came-from" target="_blank">https://openai.com/index/where-the-goblins-came-from</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Where the goblins came from 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Building the compute infrastructure for the Intelligence Age</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI scales Stargate to build the compute infrastructure powering AGI, adding new data center capacity to meet growing AI demand.</p> <p>链接：<a href="https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age" target="_blank">https://openai.com/index/building-the-compute-infrastructure-for-the-intelligence-age</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Building the compute infrastructure for the Intelligence Age 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/realtime-voice-component</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：openai/realtime-voice-component recently updated repository.</p> <p>链接：<a href="https://github.com/openai/realtime-voice-component" target="_blank">https://github.com/openai/realtime-voice-component</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/realtime-voice-component 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/jukebox</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Code for the paper "Jukebox: A Generative Model for Music"</p> <p>链接：<a href="https://github.com/openai/jukebox" target="_blank">https://github.com/openai/jukebox</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">比起表面参数，openai/jukebox 更需要观察它是否在推理质量、检索效果或可用性上带来真实改进。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>karpathy/reader3</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub karpathy</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Quick illustration of how one can easily read books together with LLMs. It's great and I highly recommend it.</p> <p>链接：<a href="https://github.com/karpathy/reader3" target="_blank">https://github.com/karpathy/reader3</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">karpathy/reader3 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-quickstarts</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：A collection of projects designed to help developers quickly get started with building deployable applications using the Claude API</p> <p>链接：<a href="https://github.com/anthropics/claude-quickstarts" target="_blank">https://github.com/anthropics/claude-quickstarts</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">如果 anthropics/claude-quickstarts 能减少集成成本、维护负担和迁移阻力，它才有进入生产栈的价值。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Cybersecurity in the Intelligence Age</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI outlines a five-part action plan for strengthening cybersecurity in the Intelligence Age, focused on democratizing AI-powered cyber defense and protecting critical systems.</p> <p>链接：<a href="https://openai.com/index/cybersecurity-in-the-intelligence-age" target="_blank">https://openai.com/index/cybersecurity-in-the-intelligence-age</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">从 Cybersecurity in the Intelligence Age 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/claude-agent-sdk-typescript</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/claude-agent-sdk-typescript recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/claude-agent-sdk-typescript" target="_blank">https://github.com/anthropics/claude-agent-sdk-typescript</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/claude-agent-sdk-typescript 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>openai/whisper</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub openai</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：Robust Speech Recognition via Large-Scale Weak Supervision</p> <p>链接：<a href="https://github.com/openai/whisper" target="_blank">https://github.com/openai/whisper</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">openai/whisper 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>anthropics/anthropic-sdk-python</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：GitHub anthropics</p> <p>标签：#github_orgs #extended</p> <p>作者： </p> <p>原文：anthropics/anthropic-sdk-python recently updated repository.</p> <p>链接：<a href="https://github.com/anthropics/anthropic-sdk-python" target="_blank">https://github.com/anthropics/anthropic-sdk-python</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">对 anthropics/anthropic-sdk-python 来说，更值得判断的是它会不会进入团队默认工具链，而不是短期讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>OpenAI models, Codex, and Managed Agents come to AWS</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI GPT models, Codex, and Managed Agents are now available on AWS, enabling enterprises to build secure AI in their AWS environments.</p> <p>链接：<a href="https://openai.com/index/openai-on-aws" target="_blank">https://openai.com/index/openai-on-aws</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">更值得关注的是 OpenAI models, Codex, and Managed Agents come to AWS 是否真正改变产品落地、工程效率、分发格局或平台控制力，而不只是制造声量。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Our commitment to community safety</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：Learn how OpenAI protects community safety in ChatGPT through model safeguards, misuse detection, policy enforcement, and collaboration with safety experts.</p> <p>链接：<a href="https://openai.com/index/our-commitment-to-community-safety" target="_blank">https://openai.com/index/our-commitment-to-community-safety</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 Our commitment to community safety，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>RT by @karpathy: New work with @AlecRad and @DavidDuvenaud: Have you ever dreamed of talking to someone from the past?</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：X Andrej Karpathy</p> <p>标签：#x_profiles #extended</p> <p>作者： </p> <p>原文：New work with @AlecRad and @DavidDuvenaud Have you ever dreamed of talking to someone from the past? Introducing talkie, a 13B model trained only on pre-1931 text. Vintage models should help us to understand how LMs generalize (e.g., can we teach talkie to code?). Thread: Video</p> <p>链接：<a href="https://twitter.com/status_effects/status/2048878495539843211" target="_blank">https://twitter.com/status_effects/status/2048878495539843211</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">RT by @karpathy: New work with @AlecRad and @DavidDuvenaud:... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>OpenAI available at FedRAMP Moderate</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：OpenAI Blog</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：OpenAI is available at FedRAMP Moderate authorization for ChatGPT Enterprise and the OpenAI API, enabling secure AI adoption for U.S. federal agencies.</p> <p>链接：<a href="https://openai.com/index/openai-available-at-fedramp-moderate" target="_blank">https://openai.com/index/openai-available-at-fedramp-moderate</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">围绕 OpenAI available at FedRAMP Moderate，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Sightings</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Simon Willison</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：/elsewhere/sightings/ I have a new camera (a Canon R6 Mark II) so I'm taking a lot more photos of birds. I share my best wildlife photos on iNaturalist and based on yesterday's successful prototype I decided to add those to my blog. I built this feature on my phone using Claude Code for web, as an extension of my beats system for syndicating external content. Here's the PR and prompt. As with my other forms of incoming syndicated content sightings show up on the homepage, the date archive pages, and in site search results. I back-populated over a decade of iNaturalist sightings, which means you that if you search for lemur you'll see my lemur photos from Madagascar in 2019! Tags: blogging photography wildlife ai inaturalist generative-ai llms ai-assisted-programming claude-code</p> <p>链接：<a href="https://simonwillison.net/2026/May/2/sightings/#atom-everything" target="_blank">https://simonwillison.net/2026/May/2/sightings/#atom-everything</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Sightings 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>AI Engineer World's Fair Autoresearch, Memory, World Models, Tokenmaxxing, Agentic Commerce, and Vertical AI Call for Speakers</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Latent Space</p> <p>标签：#ai_engineering_blogs #core</p> <p>作者： </p> <p>原文：a quiet day lets us make a call for speakers!</p> <p>链接：<a href="https://www.latent.space/p/ainews-ai-engineer-worlds-fair-autoresearch" target="_blank">https://www.latent.space/p/ainews-ai-engineer-worlds-fair-autoresearch</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">[AINews] AI Engineer World's Fair — Autoresearch, Memory, Wo... 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Most Companies Aren't Anywhere Near Ready for AI</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://twitter.com/DanielMiessler/status/2050666594188304484 Comments URL: https://news.ycombinator.com/item?id=47999238 Points: 1 Comments: 0</p> <p>链接：<a href="https://twitter.com/DanielMiessler/status/2050666594188304484" target="_blank">https://twitter.com/DanielMiessler/status/2050666594188304484</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Most Companies Aren't Anywhere Near Ready for AI 更值得从实际采用价值来判断，而不是只看它有没有制造新的讨论热度。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>Show HN: MegaLLM Universal LLM client for any OpenAI-compatible API</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://megallm.netlify.app/ Comments URL: https://news.ycombinator.com/item?id=47999114 Points: 1 Comments: 0</p> <p>链接：<a href="https://megallm.netlify.app/" target="_blank">https://megallm.netlify.app/</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">Show HN: MegaLLM – Universal LLM client for any OpenAI-compa... 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。</span></div> </div> </div> </div> <div class="brief-section"> <div class="brief-title"> <h3>How many e's are in the word seventeen [video] (AI hallucination)</h3> </div> <div class="brief-columns"> <div class="left"> <p class="meta">来源：Hacker News Newest</p> <p>标签：#research_community #extended</p> <p>作者： </p> <p>原文：Article URL: https://www.youtube.com/shorts/nks72LuZO20 Comments URL: https://news.ycombinator.com/item?id=47999082 Points: 1 Comments: 1</p> <p>链接：<a href="https://www.youtube.com/shorts/nks72LuZO20" target="_blank">https://www.youtube.com/shorts/nks72LuZO20</a></p> </div> <div class="right"> <div class="opinion-line"><span class="opinion-label">观点：</span><span class="opinion-text">How many e's are in the word seventeen [video] (AI hallucina... 本身是信号而不是结果，真正要看的是它把哪些工程和产品判断推向前沿。</span></div> </div> </div> </div> Sat, 02 May 2026 00:00:00 +0000 //2026/05/02/AI-News-Daily-Issue-63.html //2026/05/02/AI-News-Daily-Issue-63.html AI-Daily Model Andrej-Karpathy