Yann LeCun
Re @deanwball The "ungovernability" (and openness) of Linux and the Internet is precisely what has made their success.
The same will be true of open weight AI foundation models.
Vercel
this is who runs this account
This is a sick exploration of chess history.
The Replit community is ChessMaxxing
Daniel Innovate: My new Chess Experience is in beta:
https://chess-ai.app
The reason I want this btw is spending an entire day staring at screens trying to manage agents fries my brain.
Much better to be walking around outside talking to the agents "on the phone" giving them work to do and having them update me on status via voice.
Can't wait for the first lab to ship this.
Peter Yang: ChatGPT Live and Codex are two incredible products that don’t talk to each other.
This is @OpenAI's biggest missed opportunity imo.
I went on a walk with ChatGPT Live and asked it to pull up my Google Doc. It said it couldn’t.
I then manually triggered the Documents plugin to
Kenneth Roth
With Trump deeply unpopular, Republicans face a possible drubbing in congressional races in November. So Trump seems intent on laying a predicate that could explain away a defeat and potentially justify direct intervention aimed at changing the results. https://trib.al/NCwDKqc
Together AI
We analyzed Kimi K3 vs. Claude Fable 5 for software engineering tasks using DeepSWE.
Kimi K3 gets you the same performance as Fable 5 at ~35% of the price, and it actually pulls ahead at higher pass@k's. More insights in the thread!
Zain: Deepdive: Kimi K3 vs. Fable 5 on software engineering/DeepSWE tasks.
The open frontier isn't 6 months behind anymore!
full deep-dive 👇(1/n)🧵
Trump Lie Tracker (Commentary Account)
BREAKING: The last month has cemented Trump's status as the most corrupt president in American history. It's not even close.
CNN found that Trump’s money managers made at least 44 stock purchases in 21 companies within a week before he praised those companies, their executives, or their products on Truth Social.
CNBC reported that Trump’s account bought up to $5 million in Axon stock just two weeks before ICE sought a $220 million Taser contract. Axon also sells Skydio police drones now at the center of alarming revelations about police surveillance and exposed footage that violates the privacy rights of American citizens.
Meanwhile, Trump’s media company is pitching Wall Street firms $100,000-a-month access to his market-moving posts faster than the public can receive them.
Trump promised to drain the swamp. Instead, he has turned the presidency into a personal stock tip, a vehicle for profiting off government contracts, and a paywalled data feed for Wall Street.
This is corruption and self-enrichment in plain sight. You should be outraged.
Peter Yang
Legendary episode dropping this weekend featuring the one and only @trq212, who showed me his AI video workflows and much more.
📌 Subscribe to get it on Sunday: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
Legendary episode dropping this weekend featuring the one and only @trq212, who showed me his AI video workflows and much more.
📌 Subscribe to get it on Sunday: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
Ship like @shadcn
shadcn: @josefbender_ Excellent idea. And done. You can try it on the site.
Codex browser use has finally been defeated.
Peter Yang
Codex browser use has finally been defeated.
🆓 Sandbox data for downloads.
Happy Friday! Time to ship more agents
Vercel Developers: Data transfer into Vercel Sandbox is now free.
▪︎ Clone Git repos and install packages without data charges
▪︎ Pull artifacts or datasets from external sources at no cost
https://vercel.com/changelog/data-downloaded-by-vercel-sandbox-is-now-free
Y Combinator
http://ycombinator.com/apply
Kenneth Roth
"Trump called for fixing vulnerabilities in electronic voting system" but he "has spent…his second term dismantling the election protections built up over previous years." He doesn't want a fair system. He wants an excuse for Republicans' likely loss. https://trib.al/IvVhRWC
Kenneth Roth
The 270+ pages of "evidence" released by Trump show China in 2020 "considered modest attempts to influence opinion in the US, and downloaded publicly available voter rolls from several states, but never manipulated a single voting machine or ballot." https://trib.al/NUKMKyV
Chamath Palihapitiya
The future is open source. We need to embrace it and get on with it.
Imagine if America closed the door on open source.
We would explicitly be forcing American companies to pay $26-56 per 1MM tokens for the same intelligence their adversaries/competitors around the world would pay $0.50-1 for.
This is economically unsustainable unless AI is bullshit.
If it’s the supposed through line of all future economic activity that we’ve been told it is, we can’t handicap America at such a meaningful cost disadvantage.
This can also be viewed thru a military lens and not just a simple economic one as well.
Letting our adversaries attack us for $0.50 per 1MM tokens while we spend $26-56 per 1MM tokens to defend ourselves is equally ruinous.
It’s the Cold War Soviet collapse in reverse.
David Sacks: This is *exactly* what I predicted would happen. I said Chinese models would have advanced cyber capabilities within a matter of months and the only thing to do about it was to use AI-powered cyberdefense to protect our systems. Trying to gatekeep models doesn’t work.
Carlos E. Perez
http://x.com/i/article/2078418167546691584
Daniel Jeffries
You either believe in centralization of power or broad diffusion of power.
In one worldview, there are some bad people in the world we've got to restrict everyone to stop them.
In the other, you trust most people to use their kitchen knives to cut vegetables and punish the few bad people who stab someone.
One is the path of freedom. The other is the path of oligarchy and servitude.
One leads to flourishing, vibrance and dynamism. The other leads to stagnation, bloat and decline.
Choose wisely.
Daniel Jeffries
TLDR, not a single prediction of the AI safety folks has been correct and actual more mundane, everyday problems in the real world are all completely missed (and fixable with basic rules and engineering and time) while we continue to focus on the big fake imaginary problems of the future that are surely coming real soon to turn us all into paperclips!
We promise! Just trust us!
Didn't you see this scary video or read this astroturfed article I published in Forbes from my two person non-profit?
Now listen, what we've got to do to make the world a safe place and a utopia for us all is to create a brutal authoritarian centralized world regime to control everyone, spy on everyone, put tracking and kill chips in GPUs!
Every one of these kids needs to get the fuck off the Addy and the weed, switch off their computers and go touch grass.
After that volunteer at a homeless shelter and do some actual good for a real problem in the real world.
Joshua Saxe: AI safety from GPT-2 to Kimi K3
Imagine a village with a wizard who one day emerges from his cave with the following tale for his fellow villagers:
“Long poor, I will lead our village to prosperity by producing a series of ever more powerful magic wands as long as we're willing
Zane Hengsperger
YC bookface has been shipping a ton of products
i use it every day now
Sar Haribhakti
.@willpoffwebster: "If we feel like we've run out of land, like in a place like San Francisco, it's actually really easy to create more land for housing. You just build another house on top of the existing house. It's a solved problem. We've done it for hundreds of years. We just need to make it legal to do it again. It was legal 100 years ago, and then we invented zoning, and we made it illegal again."
MTS: IFP infrastructure policy director Will Poff-Webster on the housing shortage:
"We have the biggest shortage of smaller, cheaper houses in America. If you want to build a huge McMansion in the exurbs, it's not hard to do that. The problem is those houses aren't affordable to a
AI Engineer
🗽 AI Engineer returns to NYC!
Oct 12-14, a beautiful time to visit.
Tickets go on sale soon, but the Call for Speakers is now open!
https://ai.engineer/cfp
If you do leading AI work in NYC, especially for AI x Finance, speaker applications for AIE NYC 2026 opened today.
For mainstage keynotes, we are actively soliciting the top AI native companies and AI transformations in:
- Investment Banking
- Commercial Banking
- Consumer Banking
- Hedge Funds
- Private Equity
- Financial Data
- Insurance
- Venture Capital
- Agentic Commerce
- Accounting
- Payments
- Compliance
- Mid/Back Office
If you are NOT in Finance, don't worry, we will still be running our regular AI Engineering and AI Leadership tracks, covering all the latest in coding, generative media, evals, infra, and more. But if you want to meet some of the largest AI native banks and financial institutions in the world, this will be the #1 place to be this fall!
One of my biggest takeaways from chatting with @trq212 is how much effort he puts into understanding what AI is doing.
It’s easy to let AI build something magical and feel like you’re making progress. But if you want to get better, you need to read its output, identify your unknowns, and fill those gaps.
For example, Thariq sometimes asks Claude to generate an HTML report explaining what it changed and then quiz him to make sure he understands it.
This is just one of many things I learned from Thariq in our interview.
📌 Subscribe to get the full episode tomorrow: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
Peter Yang
One of my biggest takeaways from chatting with @trq212 is how much effort he puts into understanding what AI is doing.
It’s easy to let AI build something magical and feel like you’re making progress. But if you want to get better, you need to read its output, identify your unknowns, and fill those gaps.
For example, Thariq sometimes asks Claude to generate an HTML report explaining what it changed and then quiz him to make sure he understands it.
This is just one of many things I learned from Thariq in our interview.
📌 Subscribe to get the full episode tomorrow: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
California needs to embrace hard tech instead of whatever BS is happening right now
M. Nolan Gray 🥑: In a parallel world with better California governance, the San Francisco Bay Area or Greater Los Angeles would have been obvious candidates for the high-tech manufacturing boom. Instead, the American Pearl River Delta seems to be happening here.
martin_casado
"Open-weight models are inherently decelerationist"
.... this is a grossly incorrect statement with no supporting arguments or logic that is counter to the long arc of learnings of the industry over the last 50 years.
What a stupid thing to say.
Dean W. Ball: Some observations on Kimi:
1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also
martin_casado
Do you know what open weights model actually decelerate? The formation and power of oligopolies.
You know, those things which really do stifle innovation.
Dean W. Ball: Some observations on Kimi:
1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also
martin_casado
Do you know what open weights model actually decelerate? The formation and power of oligopolies.
You know, those things which really do stifle innovation.
Dean W. Ball: Some observations on Kimi:
1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also
My 8 year old and I want that basketball so we are cooking
ChatGPT: Create something worth sharing with ChatGPT sites.
Share your site with us for a chance to win a limited-edition swag box, including the sold out ChatGPT basketball 🏀
Giveaway closes this Sunday at 11:59PT, winner will be chosen randomly.
Happy building.
Sar Haribhakti
Having the common sense to make it legal to build everything has been the winning move for Texas
Ashlee Vance: So, California had the chance to build the largest shipyard in the US but couldn't be bothered to deal with it and now it's going to Texas.
How many times are we going to do this, @GavinNewsom?
Melanie Mitchell
In the off chance than anyone cares what I think on this topic:
(1) the open-source software movement has been enormously beneficial to socity
(2) open-weight (& better, open-data) LLMs will be essential for understanding this technology & for it to be beneficial to society.
I think AI psychosis (trying to make sure your agents are working all the time) is probably a worse addiction than social media.
The term "AGI" has aged very poorly. AI can't possibly me more different from human intelligence. It's *far better* than human intelligence, for most economically-relevant tasks.
But that doesn't make AIs better than humanity. There are wonderful things humans can do that these superintelligences (the better term) can't.
For one, we care-for and care-about other humans in a powerful and inherent way, that we then teach machines to emulate. We must keep that as the top priority in everything we do. We have to be as pro-human-life as it gets.
AIs can replace tasks you do, but they can't replace the proverbial you. This is why they horribly suck at writing! Even when given a corpus of your writing and 200 Skills and 50 subagents running in gRaPh LoOps, they produce robotic prose… because they are robots.
I know one definitive way that people can become irrelevant. They stop being themselves: lose their identity, delegate all their writing, their unique thoughts, the creative ways in which they can use the machines. Opting out of 'weighing in'.
I think we're safe though. There's nothing more distasteful right now than AI replies in social media or (🟢 T H I S T H I N G ) in landing pages. Quality and humanity will prevail.
Sarrah Bellus
Every time Donald Trump talks about election fraud, remind him of January 2, 2021.
The recording of him pressuring Georgia officials to “find” 11,780 votes after he lost the 2020 election.
The man who tried to overturn an election he lost has no moral authority to lecture anyone about election integrity.
England is starting without Kane or Bellingham...what?
Sasha Malysheva
latent actions are having a moment, especially in robotics: instead of predicting a robot's actual joint commands or game controller inputs, you learn a compact latent code that just explains what changed between two frames, no action labels needed anywhere in training
Genie (Bruce et al, DeepMind 2024, http://arxiv.org/abs/2402.15391) was the clean proof of concept: an inverse model looks at consecutive frames, infers a small discrete latent action that explains the transition, and that learned action space alone is enough to make the generated video controllable frame by frame, despite the model never seeing a single ground-truth action label (the same latent space also turns out to transfer, you can use it to train imitation agents on videos that were never annotated for any specific robot)
(and in the same research cluster: Video as the New Language for Real-World Decision Making, http://arxiv.org/abs/2402.17139 by @sherryyangML et al., argues the broader case for video itself as the general interface, not just for latent actions specifically)
the harder problem is generalizing latent actions beyond narrow domains: Genie was trained on platformer game footage with one consistent embodiment and clean signal, while a new paper from FAIR (@garridoq_, Nagarajan, Terver, Ballas, @ylecun, Rabbat, http://arxiv.org/abs/2601.05230) pushes latent action learning onto actual in-the-wild internet video, which means no shared embodiment across clips and a lot more environmental noise to filter out before the latent action means anything consistent
why robotics specifically cares: action labels require teleoperation or motion capture, which doesn't scale, however unlabeled video of things happening in the world is nearly infinite, so if the latent action space generalizes, you get to train on all of it instead of just what you paid someone to label
Clifford Sosin
http://x.com/i/article/2078590656457625600
Clifford Sosin
http://x.com/i/article/2078590656457625600
Garry Kasparov
"Russia hoax" is as strong an indicator of a being gullible idiot or partisan propagandist as "vaccines are bad", and both ignore overwhelming evidence.
Tommy Vietor: Trump released documents that confirm Russia tried to interfere in the 2020 election to hurt Biden. It says Russia tried to "orchestrate a high-profile corruption scandal" and push narratives about "voter fraud resulting from mail-in balloting"
This is the weirdest match
Peter Yang: Lmao why is France playing so badly
Andrew Blumson
I shared a game I was building after 2 or 3 prompts. This is the same game a few hours later - I'd say now around 50% complete.
Still to do - add road objects; cars, buses, taxis, etc.
But the play mechanics and physics are solid because of the work I did in prompts 1, 2 and 3. Game plays lovely. I'll make it look nice at the end.
@Replit #GameDev #BuildInPublic #VibeCoding
People saying this is the greatest World Cup match of this tournament...smh
people hating on europe miss that they actually have some of the top ai engineers in the world if you actually to elicit the right ones.
we're basically running the most competitive global arena for ai talent (as measured by talks and workshops), will be v interesting to see how this goes as WF enters the eval window
x.com
