Guillermo Rauch@rauchg·9小时前Generative UI for conversational agents. Combining @aisdk with Chat SDK and 𝚓𝚜𝚘𝚗-𝚛𝚎𝚗𝚍𝚎𝚛, you can go straight from LLM to UI on platforms like Slack and Teams. I believe this will become the dominant Human-Agent Interaction (HAI) paradigm: https://vercel.com/changelog/json-render-and-chat-sdk-agents翻译
thsottiaux@thsottiaux·13小时前R to @thsottiaux: More things to come, join our Cyber at the Frontier livestream on November 12, 10am PT: https://openai.com/cyber-at-the-frontier翻译
OpenAI@OpenAI·13小时前We partnered with @fmf_org and its member firms on a new framework to help identify when bypassing an AI system's safeguards could enable serious cyber harm, and prioritize stronger protections. "...This framework is an important step toward a common methodology for evaluating and comparing the severity of cyber jailbreaks, including the capabilities they unlock and how readily they can be used. Shared approaches like this can help the AI and security communities build a clearer evidence base and prioritize improvements to safeguards." - OpenAI's Sasha Baker, Head of National Security Policy https://www.frontiermodelforum.org/updates/introducing-the-fmf-jailbreak-rating-framework-for-frontier-cyber-threats/翻译
OpenAI@OpenAI·14小时前A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion. Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity. https://openai.com/index/measuring-how-ai-agents-handle-wet-lab-science/翻译
Guillermo Rauch@rauchg·15小时前I unironically agree with @bcherny’s take that “coding is solved, but bugs are not.” The deeper meaning is that you can in theory code *anything* now. You can burn infinite tokens refining and optimizing. Including optimizations that no one will care about 💀 With Vercel you can build a different kind of recursive improvement loop. You capture traces from production, you then optimize and course-correct based on how your software is being used IRL. This is not a new idea. The best engineers I’ve ever worked with never jumped into coding (now prompting) right away. You instrument the system, collect data, understand distributions, user journeys, and then put fingers on keyboard. You can ask your agent today to look at your traces!翻译
levelsio@levelsio·16小时前Also funny how we repeat the cycles 9 years ago I made http://mute.life because the social justice and politics on here got so bad It was a site with words you could mute and with the bookmarklet import into your Twitter account to make your timeline usable again Now we have the same issue but with AI replies!翻译
levelsio@levelsio·16小时前One of my biggest motivations to make web games (the original motivation for my fly-pieter-com game) or port existing games to the web is how incredibly slow the loading times of modern games have become I have an RTX 5090 w/ 32GB GDDR7 liquid cooled maxed out gaming PC with 128GB DDR5 RAM, a 16-core 32-thread AMD Ryzen 9 9950X3D and a Corsair 4TB NVMe SSD And it still takes almost 3 minutes just to load Flight Simulator, and that's after loading it a few minutes before already to install the updates You can call me impatient, but you used to just put you'd run a game and you'd start playing within 30 seconds or instantly! That's how it should be I think!翻译
levelsio@levelsio·16小时前iOS 27 has broken the amazingly useful Low Light feature (inside Zoom) combined with Smart Invert If anyone works at @Apple, or knows anyone there, it'd be amazing to get this working again! Bug report and prompt for Apple staff to copy paste into their clanker to fix it: "## Bug: Zoom "Low Light" filter + Smart Invert makes the screen brighter instead of darker (regression in iOS 27) ### Summary Before iOS 27, turning on Smart Invert together with the Zoom filter "Low Light" made the screen very dark. Many users rely on this to read at night, like a Kindle. In iOS 27 the same combination makes the screen noticeably BRIGHTER than Smart Invert alone: the background looks washed-out light grey and the text is full white. ### Steps to reproduce 1. Settings → Accessibility → Display & Text Size → turn on Smart Invert. 2. Settings → Accessibility → Zoom → turn on Zoom, set Zoom Filter to "Low Light", Zoom Region to Full Screen, and zoom level 1.0x (or the lowest possible). 3. Open a white page, e.g. Safari or Books. 4. Compare the screen with Low Light on and off. ### Expected (iOS 26 and earlier) - White background → near black - Black text → dim grey - Result: darker than Smart Invert alone ### Actual (iOS 27) - White background → light grey (~70%) - Black text → full white - Result: brighter and lower contrast than Smart Invert alone ### Likely root cause The order of the two filters seems to have been swapped in the display/compositing pipeline. - Old order: Smart Invert → Low Light (invert, then dim) - Output = 0.3 × (1 − x), which is always dark - New order: Low Light → Smart Invert (dim, then invert) - Output = 1 − 0.3 × x, which is mostly bright, because inverting a dimmed image turns the dimming into brightening The washed-out grey look is the typical sign of dimming before inverting. My guess is that a refactor in iOS 27 moved Smart Invert into a final full-screen pass, or moved the Zoom filter earlier in the chain, so Low Light now runs underneath the inversio翻译
levelsio@levelsio·17小时前Eerily quit in my replies and suddenly no AI bot replies anymore 😂 I think some people at least got the message?翻译
levelsio@levelsio·17小时前The worst part about AI replies it's a negative cost to all of us on here You squeeze out people who actually put effort to interact with others on here For some dumb growth strategy that doesn't work anyway TL:DR: if you're doing AI replies, you're an asshole翻译
levelsio@levelsio·17小时前I will publish my full list of 10,000+ blocked AI spammer accounts soon so everyone can block them easily tooo If you write AI replies under my tweets you will now permanently destroy your X account Because you'll very quickly end up in a public data set of AI spammers that people will use to block them Better turn those AI reply bots off today翻译
levelsio@levelsio·17小时前Great list of AI spammers Ask Claude in Chrome to block everyone in this list翻译
TheRundownAI@TheRundownAI·18小时前R to @TheRundownAI: Read more: https://tech.therundown.ai/p/big-tech-sues-to-save-the-feed翻译
TheRundownAI@TheRundownAI·18小时前Top stories in tech today: - Meta, Google, TikTok sue California - Musk clears Cybertruck off Giga Texas - Google drops $17B on Wyoming data center - Sony revives its Walkman headphones翻译
soltwagner@soltwagner·18小时前Create tasks easily in the new https://Woday.app 1.2 update, just launched 🙌 Which option is your favorite for creating tasks? 👀翻译
levelsio@levelsio·18小时前I will collect all accounts I blocked for AI spam recently and submit them to @dillon_mulroy's list too I will also send my entire list to @X staff so they can suspend the accounts If you post AI replies under my tweets I will now go nuclear on you Only way to keep this platform usable翻译
levelsio@levelsio·22小时前Crypto bros go to Paraguay Forced to open bank account to have ties Bank fails Ironically proving their point翻译
levelsio@levelsio·22小时前It's so nice to read this I've worked with Tommy for almost 12 years, which is a significant portion of my life and much longer than I've known most people these days I learned so much about Linux, sysadmin and performance optimization over that time. I'm still impressed how much impact a single server has had on the world Couldn't have done it without the batshit insane ideas from @levelsio of course Thank you for your service Tommy!! ❤️翻译
levelsio@levelsio·23小时前I did the same a few months ago I shut down my 13-year old $300/mo Linode VPS server that hosted all my 100+ sites for over a decade! That Linode VPS called Tommy is where everything started for me, I knew to build startups I needed a server so I just made one and SSH'd into it and started uploading PHP files of little startups I made After 1.5 years my first sites started taking off and I started making money on the internet with Tommy! For almost a decade I'd still FTP my files by dragging them every time I made an edit to the server, only around 2020 I switched to Github because @daniellockyer said it'd be better and because regularly the site would break at halfway because people would load the site while I was in the middle transferring index.php over FTP 😂 I didn't have multiple VPS servers, I had just one for ALL my sites and projects (I'd say upwards of 100), but it worked mostly fine, and I regularly posted about how it was all on one server which made Twitter/X go crazy "how do you scale???", it scaled fine, served 300 million people per year and 115,000 paying customers in its lifetime! Back around 2013, Linode was the "cool" hoster on Hacker News, but a decade later, the spirit of Linode seemed to have disappeared, it felt like it was in maintenance mode and not really loved for anymore. When it was acquired by Akamai in 2022, me and @daniellockyer knew we'd have to leave because acquired companies generally go to shit I saw @DHH posting about @Hetzner_Online so I started making little VPS servers there to test it, first with @pieter and it worked really great. Then slowly I started migrating each site over to Hetzner, with the most critical sites like Photo AI and Nomads last. I also started separating sites on to different VPS servers, isolating them, which now in this new AI cybersecurity time makes security sense. It was never an issue with performance though to have 100+ sites on one VPS, like literally never at all Then a few months ago I had to f翻译
thsottiaux@thsottiaux·1天前R to @thsottiaux: Formal proofs of tens of millions of lines of code are not a ridiculous thing anymore and within the realm of what should be considered when designing new systems.翻译
thsottiaux@thsottiaux·1天前I can confidently say that humans were not very good at writing secure infrastructure. The endstate will be hypersecure infrastructure and software for all, but we're going to go through some turbulence before we reach that.翻译
theworldlabs@theworldlabs·1天前We're in a computer graphics renaissance Drawings, photos, renders are all something world models can "see into", world vertigo翻译
steipete@steipete·1天前Is my perception off or is Apple just lying here? Anytime I build for or use Safari, it’s slower, has fewer features, weird bugs and plenty websites just don’t work. It’s a fight to get even somewhat close to Chrome’s performance.翻译
levelsio@levelsio·1天前Many of my friends like @csonotes and @NorbertDragan devoutly believe we have to go outside more and I think AI will help us to :D翻译
levelsio@levelsio·1天前Same with AI computering has become boring and with that being indoors has become boring翻译
Guillermo Rauch@rauchg·1天前① Get an X sub ② Build with Bot / Cursor ③ Deploy to Vercel, scales from 0 to 🌙 ④ Launch and distribute on X ⑤ Monetize with X Money / Creators The loop is nearly complete ♾️翻译
levelsio@levelsio·1天前Rate my painting skills We had to weatherproof the outdoor sauna because it's been a 1.5 year old now and otherwise it get taken over by nature itself Maybe this can be my new post-AGI job? 🙏 😬翻译
thsottiaux@thsottiaux·1天前Wild stat. We are now generate more than 4B images per week across ChatGPT and the API, up from the more than 3B per week only 8 weeks ago.翻译
Guillermo Rauch@rauchg·1天前Deploy apps in one click with @cursor_ai. • No account to create • No deployment configuration • Database provisioning included Built on Vercel for Platforms → https://vercel.com/for/platforms翻译
Guillermo Rauch@rauchg·1天前SpaceXSI 🤝 Vercel Grok and Cursor can now build, deploy, and scale https://vercel.com/for/platforms翻译
steipete@steipete·1天前I spent 20,000 dollars in API credits having different AI models play each other at Blood on the Clocktower (a Mafia or Werewolf type social deception game). The results were fascinating. Models have gotten much better over the course of this year at creative on-the-spot scheming and bluffing, and also at reasoning as a team to break apart these underground conspiracies. I go through some of the most standout games these AIs played in this video. The new models are getting pretty good. If my friends pulled off some of these moves while we were playing, I’d give them a fist bump. Fwiw, OpenAI’s newer models are much better than Anthropic’s at this game. You can see the full benchmark at the website, and I’ve also made it so that you and your friends can play again the AIs: http://clocktowerbench.com翻译
Guillermo Rauch@rauchg·1天前“Doing stuff for you over time” is the killer app of agents. Muse and Instinct consolidate memories at night, remind you of stuff in the morning, message you after N days passed since your last interaction… Vercel Schedules are like Cron on steroids. They can be defined at runtime, scale infinitely, and sit outside of the Sandbox, which means you can cost-efficiently invoke compute as needed.翻译
levelsio@levelsio·1天前P.S. This is good, I built my own expenses tracker with Claude Code, it's how I found so much stuff to save on It just proves my point that almost every industry will be entered by Anthropic, OpenAI and probably xAI too翻译
Guillermo Rauch@rauchg·1天前I had a hard time believing this… but I had https://fx.sh build a repro with 𝚊𝚕𝚙𝚒𝚗𝚎:𝟹.𝟸𝟺. ① Type in a wrong password ② Type in a > 256 chars password ③ root Alpine team has pushed a fixed apk for 𝚘𝚙𝚎𝚗𝚜𝚜𝚑-𝚜𝚎𝚛𝚟𝚎𝚛 (𝟷𝟶.𝟹_𝚙𝟷-𝚛𝟸)翻译
Guillermo Rauch@rauchg·1天前Your agent can fetch vulnerabilities found by Vercel Security Review using Vercel CLI翻译